source: src/molecules.cpp@ aa5702

Action_Thermostats Add_AtomRandomPerturbation Add_FitFragmentPartialChargesAction Add_RotateAroundBondAction Add_SelectAtomByNameAction Added_ParseSaveFragmentResults AddingActions_SaveParseParticleParameters Adding_Graph_to_ChangeBondActions Adding_MD_integration_tests Adding_ParticleName_to_Atom Adding_StructOpt_integration_tests AtomFragments Automaking_mpqc_open AutomationFragmentation_failures Candidate_v1.5.4 Candidate_v1.6.0 Candidate_v1.6.1 ChangeBugEmailaddress ChangingTestPorts ChemicalSpaceEvaluator CombiningParticlePotentialParsing Combining_Subpackages Debian_Package_split Debian_package_split_molecuildergui_only Disabling_MemDebug Docu_Python_wait EmpiricalPotential_contain_HomologyGraph EmpiricalPotential_contain_HomologyGraph_documentation Enable_parallel_make_install Enhance_userguide Enhanced_StructuralOptimization Enhanced_StructuralOptimization_continued Example_ManyWaysToTranslateAtom Exclude_Hydrogens_annealWithBondGraph FitPartialCharges_GlobalError Fix_BoundInBox_CenterInBox_MoleculeActions Fix_ChargeSampling_PBC Fix_ChronosMutex Fix_FitPartialCharges Fix_FitPotential_needs_atomicnumbers Fix_ForceAnnealing Fix_IndependentFragmentGrids Fix_ParseParticles Fix_ParseParticles_split_forward_backward_Actions Fix_PopActions Fix_QtFragmentList_sorted_selection Fix_Restrictedkeyset_FragmentMolecule Fix_StatusMsg Fix_StepWorldTime_single_argument Fix_Verbose_Codepatterns Fix_fitting_potentials Fixes ForceAnnealing_goodresults ForceAnnealing_oldresults ForceAnnealing_tocheck ForceAnnealing_with_BondGraph ForceAnnealing_with_BondGraph_continued ForceAnnealing_with_BondGraph_continued_betteresults ForceAnnealing_with_BondGraph_contraction-expansion FragmentAction_writes_AtomFragments FragmentMolecule_checks_bonddegrees GeometryObjects Gui_Fixes Gui_displays_atomic_force_velocity ImplicitCharges IndependentFragmentGrids IndependentFragmentGrids_IndividualZeroInstances IndependentFragmentGrids_IntegrationTest IndependentFragmentGrids_Sole_NN_Calculation JobMarket_RobustOnKillsSegFaults JobMarket_StableWorkerPool JobMarket_unresolvable_hostname_fix MoreRobust_FragmentAutomation ODR_violation_mpqc_open PartialCharges_OrthogonalSummation PdbParser_setsAtomName PythonUI_with_named_parameters QtGui_reactivate_TimeChanged_changes Recreated_GuiChecks Rewrite_FitPartialCharges RotateToPrincipalAxisSystem_UndoRedo SaturateAtoms_findBestMatching SaturateAtoms_singleDegree StoppableMakroAction Subpackage_CodePatterns Subpackage_JobMarket Subpackage_LinearAlgebra Subpackage_levmar Subpackage_mpqc_open Subpackage_vmg Switchable_LogView ThirdParty_MPQC_rebuilt_buildsystem TrajectoryDependenant_MaxOrder TremoloParser_IncreasedPrecision TremoloParser_MultipleTimesteps TremoloParser_setsAtomName Ubuntu_1604_changes stable
Last change on this file since aa5702 was aa5702, checked in by Frederik Heber <heber@…>, 17 years ago

corrected molecule::VerletForceIntegration() and analyzer creates data and plot file for shielding values

  • Property mode set to 100644
File size: 241.8 KB
Line 
1/** \file molecules.cpp
2 *
3 * Functions for the class molecule.
4 *
5 */
6
7#include "molecules.hpp"
8
9/************************************* Other Functions *************************************/
10
11/** Determines sum of squared distances of \a X to all \a **vectors.
12 * \param *x reference vector
13 * \param *params
14 * \return sum of square distances
15 */
16double LSQ (const gsl_vector * x, void * params)
17{
18 double sum = 0.;
19 struct LSQ_params *par = (struct LSQ_params *)params;
20 Vector **vectors = par->vectors;
21 int num = par->num;
22
23 for (int i=num;i--;) {
24 for(int j=NDIM;j--;)
25 sum += (gsl_vector_get(x,j) - (vectors[i])->x[j])*(gsl_vector_get(x,j) - (vectors[i])->x[j]);
26 }
27
28 return sum;
29};
30
31/************************************* Functions for class molecule *********************************/
32
33/** Constructor of class molecule.
34 * Initialises molecule list with correctly referenced start and end, and sets molecule::last_atom to zero.
35 */
36molecule::molecule(periodentafel *teil)
37{
38 // init atom chain list
39 start = new atom;
40 end = new atom;
41 start->father = NULL;
42 end->father = NULL;
43 link(start,end);
44 // init bond chain list
45 first = new bond(start, end, 1, -1);
46 last = new bond(start, end, 1, -1);
47 link(first,last);
48 // other stuff
49 MDSteps = 0;
50 last_atom = 0;
51 elemente = teil;
52 AtomCount = 0;
53 BondCount = 0;
54 NoNonBonds = 0;
55 NoNonHydrogen = 0;
56 NoCyclicBonds = 0;
57 ListOfBondsPerAtom = NULL;
58 NumberOfBondsPerAtom = NULL;
59 ElementCount = 0;
60 for(int i=MAX_ELEMENTS;i--;)
61 ElementsInMolecule[i] = 0;
62 cell_size[0] = cell_size[2] = cell_size[5]= 20.;
63 cell_size[1] = cell_size[3] = cell_size[4]= 0.;
64};
65
66/** Destructor of class molecule.
67 * Initialises molecule list with correctly referenced start and end, and sets molecule::last_atom to zero.
68 */
69molecule::~molecule()
70{
71 if (ListOfBondsPerAtom != NULL)
72 for(int i=AtomCount;i--;)
73 Free((void **)&ListOfBondsPerAtom[i], "molecule::~molecule: ListOfBondsPerAtom[i]");
74 Free((void **)&ListOfBondsPerAtom, "molecule::~molecule: ListOfBondsPerAtom");
75 Free((void **)&NumberOfBondsPerAtom, "molecule::~molecule: NumberOfBondsPerAtom");
76 CleanupMolecule();
77 delete(first);
78 delete(last);
79 delete(end);
80 delete(start);
81};
82
83/** Adds given atom \a *pointer from molecule list.
84 * Increases molecule::last_atom and gives last number to added atom and names it according to its element::abbrev and molecule::AtomCount
85 * \param *pointer allocated and set atom
86 * \return true - succeeded, false - atom not found in list
87 */
88bool molecule::AddAtom(atom *pointer)
89{
90 if (pointer != NULL) {
91 pointer->sort = &pointer->nr;
92 pointer->nr = last_atom++; // increase number within molecule
93 AtomCount++;
94 if (pointer->type != NULL) {
95 if (ElementsInMolecule[pointer->type->Z] == 0)
96 ElementCount++;
97 ElementsInMolecule[pointer->type->Z]++; // increase number of elements
98 if (pointer->type->Z != 1)
99 NoNonHydrogen++;
100 if (pointer->Name == NULL) {
101 Free((void **)&pointer->Name, "molecule::AddAtom: *pointer->Name");
102 pointer->Name = (char *) Malloc(sizeof(char)*6, "molecule::AddAtom: *pointer->Name");
103 sprintf(pointer->Name, "%2s%02d", pointer->type->symbol, pointer->nr+1);
104 }
105 }
106 return add(pointer, end);
107 } else
108 return false;
109};
110
111/** Adds a copy of the given atom \a *pointer from molecule list.
112 * Increases molecule::last_atom and gives last number to added atom.
113 * \param *pointer allocated and set atom
114 * \return true - succeeded, false - atom not found in list
115 */
116atom * molecule::AddCopyAtom(atom *pointer)
117{
118 if (pointer != NULL) {
119 atom *walker = new atom();
120 walker->type = pointer->type; // copy element of atom
121 walker->x.CopyVector(&pointer->x); // copy coordination
122 walker->v.CopyVector(&pointer->v); // copy velocity
123 walker->FixedIon = pointer->FixedIon;
124 walker->sort = &walker->nr;
125 walker->nr = last_atom++; // increase number within molecule
126 walker->father = pointer; //->GetTrueFather();
127 walker->Name = (char *) Malloc(sizeof(char)*strlen(pointer->Name)+1, "molecule::AddCopyAtom: *Name");
128 strcpy (walker->Name, pointer->Name);
129 add(walker, end);
130 if ((pointer->type != NULL) && (pointer->type->Z != 1))
131 NoNonHydrogen++;
132 AtomCount++;
133 return walker;
134 } else
135 return NULL;
136};
137
138/** Adds a Hydrogen atom in replacement for the given atom \a *partner in bond with a *origin.
139 * Here, we have to distinguish between single, double or triple bonds as stated by \a BondDegree, that each demand
140 * a different scheme when adding \a *replacement atom for the given one.
141 * -# Single Bond: Simply add new atom with bond distance rescaled to typical hydrogen one
142 * -# Double Bond: Here, we need the **BondList of the \a *origin atom, by scanning for the other bonds instead of
143 * *Bond, we use the through these connected atoms to determine the plane they lie in, vector::MakeNormalvector().
144 * The orthonormal vector to this plane along with the vector in *Bond direction determines the plane the two
145 * replacing hydrogens shall lie in. Now, all remains to do is take the usual hydrogen double bond angle for the
146 * element of *origin and form the sin/cos admixture of both plane vectors for the new coordinates of the two
147 * hydrogens forming this angle with *origin.
148 * -# Triple Bond: The idea is to set up a tetraoid (C1-H1-H2-H3) (however the lengths \f$b\f$ of the sides of the base
149 * triangle formed by the to be added hydrogens are not equal to the typical bond distance \f$l\f$ but have to be
150 * determined from the typical angle \f$\alpha\f$ for a hydrogen triple connected to the element of *origin):
151 * We have the height \f$d\f$ as the vector in *Bond direction (from triangle C1-H1-H2).
152 * \f[ h = l \cdot \cos{\left (\frac{\alpha}{2} \right )} \qquad b = 2l \cdot \sin{\left (\frac{\alpha}{2} \right)} \quad \rightarrow \quad d = l \cdot \sqrt{\cos^2{\left (\frac{\alpha}{2} \right)}-\frac{1}{3}\cdot\sin^2{\left (\frac{\alpha}{2}\right )}}
153 * \f]
154 * vector::GetNormalvector() creates one orthonormal vector from this *Bond vector and vector::MakeNormalvector creates
155 * the third one from the former two vectors. The latter ones form the plane of the base triangle mentioned above.
156 * The lengths for these are \f$f\f$ and \f$g\f$ (from triangle H1-H2-(center of H1-H2-H3)) with knowledge that
157 * the median lines in an isosceles triangle meet in the center point with a ratio 2:1.
158 * \f[ f = \frac{b}{\sqrt{3}} \qquad g = \frac{b}{2}
159 * \f]
160 * as the coordination of all three atoms in the coordinate system of these three vectors:
161 * \f$\pmatrix{d & f & 0}\f$, \f$\pmatrix{d & -0.5 \cdot f & g}\f$ and \f$\pmatrix{d & -0.5 \cdot f & -g}\f$.
162 *
163 * \param *out output stream for debugging
164 * \param *Bond pointer to bond between \a *origin and \a *replacement
165 * \param *TopOrigin son of \a *origin of upper level molecule (the atom added to this molecule as a copy of \a *origin)
166 * \param *origin pointer to atom which acts as the origin for scaling the added hydrogen to correct bond length
167 * \param *replacement pointer to the atom which shall be copied as a hydrogen atom in this molecule
168 * \param **BondList list of bonds \a *replacement has (necessary to determine plane for double and triple bonds)
169 * \param NumBond number of bonds in \a **BondList
170 * \param isAngstroem whether the coordination of the given atoms is in AtomicLength (false) or Angstrom(true)
171 * \return number of atoms added, if < bond::BondDegree then something went wrong
172 * \todo double and triple bonds splitting (always use the tetraeder angle!)
173 */
174bool molecule::AddHydrogenReplacementAtom(ofstream *out, bond *TopBond, atom *BottomOrigin, atom *TopOrigin, atom *TopReplacement, bond **BondList, int NumBond, bool IsAngstroem)
175{
176 double bondlength; // bond length of the bond to be replaced/cut
177 double bondangle; // bond angle of the bond to be replaced/cut
178 double BondRescale; // rescale value for the hydrogen bond length
179 bool AllWentWell = true; // flag gathering the boolean return value of molecule::AddAtom and other functions, as return value on exit
180 bond *FirstBond = NULL, *SecondBond = NULL; // Other bonds in double bond case to determine "other" plane
181 atom *FirstOtherAtom = NULL, *SecondOtherAtom = NULL, *ThirdOtherAtom = NULL; // pointer to hydrogen atoms to be added
182 double b,l,d,f,g, alpha, factors[NDIM]; // hold temporary values in triple bond case for coordination determination
183 Vector Orthovector1, Orthovector2; // temporary vectors in coordination construction
184 Vector InBondvector; // vector in direction of *Bond
185 bond *Binder = NULL;
186 double *matrix;
187
188// *out << Verbose(3) << "Begin of AddHydrogenReplacementAtom." << endl;
189 // create vector in direction of bond
190 InBondvector.CopyVector(&TopReplacement->x);
191 InBondvector.SubtractVector(&TopOrigin->x);
192 bondlength = InBondvector.Norm();
193
194 // is greater than typical bond distance? Then we have to correct periodically
195 // the problem is not the H being out of the box, but InBondvector have the wrong direction
196 // due to TopReplacement or Origin being on the wrong side!
197 if (bondlength > BondDistance) {
198// *out << Verbose(4) << "InBondvector is: ";
199// InBondvector.Output(out);
200// *out << endl;
201 Orthovector1.Zero();
202 for (int i=NDIM;i--;) {
203 l = TopReplacement->x.x[i] - TopOrigin->x.x[i];
204 if (fabs(l) > BondDistance) { // is component greater than bond distance
205 Orthovector1.x[i] = (l < 0) ? -1. : +1.;
206 } // (signs are correct, was tested!)
207 }
208 matrix = ReturnFullMatrixforSymmetric(cell_size);
209 Orthovector1.MatrixMultiplication(matrix);
210 InBondvector.SubtractVector(&Orthovector1); // subtract just the additional translation
211 Free((void **)&matrix, "molecule::AddHydrogenReplacementAtom: *matrix");
212 bondlength = InBondvector.Norm();
213// *out << Verbose(4) << "Corrected InBondvector is now: ";
214// InBondvector.Output(out);
215// *out << endl;
216 } // periodic correction finished
217
218 InBondvector.Normalize();
219 // get typical bond length and store as scale factor for later
220 BondRescale = TopOrigin->type->HBondDistance[TopBond->BondDegree-1];
221 if (BondRescale == -1) {
222 cerr << Verbose(3) << "WARNING: There is no typical bond distance for bond (" << TopOrigin->Name << "<->" << TopReplacement->Name << ") of degree " << TopBond->BondDegree << "!" << endl;
223 BondRescale = bondlength;
224 } else {
225 if (!IsAngstroem)
226 BondRescale /= (1.*AtomicLengthToAngstroem);
227 }
228
229 // discern single, double and triple bonds
230 switch(TopBond->BondDegree) {
231 case 1:
232 FirstOtherAtom = new atom(); // new atom
233 FirstOtherAtom->type = elemente->FindElement(1); // element is Hydrogen
234 FirstOtherAtom->v.CopyVector(&TopReplacement->v); // copy velocity
235 FirstOtherAtom->FixedIon = TopReplacement->FixedIon;
236 if (TopReplacement->type->Z == 1) { // neither rescale nor replace if it's already hydrogen
237 FirstOtherAtom->father = TopReplacement;
238 BondRescale = bondlength;
239 } else {
240 FirstOtherAtom->father = NULL; // if we replace hydrogen, we mark it as our father, otherwise we are just an added hydrogen with no father
241 }
242 InBondvector.Scale(&BondRescale); // rescale the distance vector to Hydrogen bond length
243 FirstOtherAtom->x.CopyVector(&TopOrigin->x); // set coordination to origin ...
244 FirstOtherAtom->x.AddVector(&InBondvector); // ... and add distance vector to replacement atom
245 AllWentWell = AllWentWell && AddAtom(FirstOtherAtom);
246// *out << Verbose(4) << "Added " << *FirstOtherAtom << " at: ";
247// FirstOtherAtom->x.Output(out);
248// *out << endl;
249 Binder = AddBond(BottomOrigin, FirstOtherAtom, 1);
250 Binder->Cyclic = false;
251 Binder->Type = TreeEdge;
252 break;
253 case 2:
254 // determine two other bonds (warning if there are more than two other) plus valence sanity check
255 for (int i=0;i<NumBond;i++) {
256 if (BondList[i] != TopBond) {
257 if (FirstBond == NULL) {
258 FirstBond = BondList[i];
259 FirstOtherAtom = BondList[i]->GetOtherAtom(TopOrigin);
260 } else if (SecondBond == NULL) {
261 SecondBond = BondList[i];
262 SecondOtherAtom = BondList[i]->GetOtherAtom(TopOrigin);
263 } else {
264 *out << Verbose(3) << "WARNING: Detected more than four bonds for atom " << TopOrigin->Name;
265 }
266 }
267 }
268 if (SecondOtherAtom == NULL) { // then we have an atom with valence four, but only 3 bonds: one to replace and one which is TopBond (third is FirstBond)
269 SecondBond = TopBond;
270 SecondOtherAtom = TopReplacement;
271 }
272 if (FirstOtherAtom != NULL) { // then we just have this double bond and the plane does not matter at all
273// *out << Verbose(3) << "Regarding the double bond (" << TopOrigin->Name << "<->" << TopReplacement->Name << ") to be constructed: Taking " << FirstOtherAtom->Name << " and " << SecondOtherAtom->Name << " along with " << TopOrigin->Name << " to determine orthogonal plane." << endl;
274
275 // determine the plane of these two with the *origin
276 AllWentWell = AllWentWell && Orthovector1.MakeNormalVector(&TopOrigin->x, &FirstOtherAtom->x, &SecondOtherAtom->x);
277 } else {
278 Orthovector1.GetOneNormalVector(&InBondvector);
279 }
280 //*out << Verbose(3)<< "Orthovector1: ";
281 //Orthovector1.Output(out);
282 //*out << endl;
283 // orthogonal vector and bond vector between origin and replacement form the new plane
284 Orthovector1.MakeNormalVector(&InBondvector);
285 Orthovector1.Normalize();
286 //*out << Verbose(3) << "ReScaleCheck: " << Orthovector1.Norm() << " and " << InBondvector.Norm() << "." << endl;
287
288 // create the two Hydrogens ...
289 FirstOtherAtom = new atom();
290 SecondOtherAtom = new atom();
291 FirstOtherAtom->type = elemente->FindElement(1);
292 SecondOtherAtom->type = elemente->FindElement(1);
293 FirstOtherAtom->v.CopyVector(&TopReplacement->v); // copy velocity
294 FirstOtherAtom->FixedIon = TopReplacement->FixedIon;
295 SecondOtherAtom->v.CopyVector(&TopReplacement->v); // copy velocity
296 SecondOtherAtom->FixedIon = TopReplacement->FixedIon;
297 FirstOtherAtom->father = NULL; // we are just an added hydrogen with no father
298 SecondOtherAtom->father = NULL; // we are just an added hydrogen with no father
299 bondangle = TopOrigin->type->HBondAngle[1];
300 if (bondangle == -1) {
301 *out << Verbose(3) << "WARNING: There is no typical bond angle for bond (" << TopOrigin->Name << "<->" << TopReplacement->Name << ") of degree " << TopBond->BondDegree << "!" << endl;
302 bondangle = 0;
303 }
304 bondangle *= M_PI/180./2.;
305// *out << Verbose(3) << "ReScaleCheck: InBondvector ";
306// InBondvector.Output(out);
307// *out << endl;
308// *out << Verbose(3) << "ReScaleCheck: Orthovector ";
309// Orthovector1.Output(out);
310// *out << endl;
311// *out << Verbose(3) << "Half the bond angle is " << bondangle << ", sin and cos of it: " << sin(bondangle) << ", " << cos(bondangle) << endl;
312 FirstOtherAtom->x.Zero();
313 SecondOtherAtom->x.Zero();
314 for(int i=NDIM;i--;) { // rotate by half the bond angle in both directions (InBondvector is bondangle = 0 direction)
315 FirstOtherAtom->x.x[i] = InBondvector.x[i] * cos(bondangle) + Orthovector1.x[i] * (sin(bondangle));
316 SecondOtherAtom->x.x[i] = InBondvector.x[i] * cos(bondangle) + Orthovector1.x[i] * (-sin(bondangle));
317 }
318 FirstOtherAtom->x.Scale(&BondRescale); // rescale by correct BondDistance
319 SecondOtherAtom->x.Scale(&BondRescale);
320 //*out << Verbose(3) << "ReScaleCheck: " << FirstOtherAtom->x.Norm() << " and " << SecondOtherAtom->x.Norm() << "." << endl;
321 for(int i=NDIM;i--;) { // and make relative to origin atom
322 FirstOtherAtom->x.x[i] += TopOrigin->x.x[i];
323 SecondOtherAtom->x.x[i] += TopOrigin->x.x[i];
324 }
325 // ... and add to molecule
326 AllWentWell = AllWentWell && AddAtom(FirstOtherAtom);
327 AllWentWell = AllWentWell && AddAtom(SecondOtherAtom);
328// *out << Verbose(4) << "Added " << *FirstOtherAtom << " at: ";
329// FirstOtherAtom->x.Output(out);
330// *out << endl;
331// *out << Verbose(4) << "Added " << *SecondOtherAtom << " at: ";
332// SecondOtherAtom->x.Output(out);
333// *out << endl;
334 Binder = AddBond(BottomOrigin, FirstOtherAtom, 1);
335 Binder->Cyclic = false;
336 Binder->Type = TreeEdge;
337 Binder = AddBond(BottomOrigin, SecondOtherAtom, 1);
338 Binder->Cyclic = false;
339 Binder->Type = TreeEdge;
340 break;
341 case 3:
342 // take the "usual" tetraoidal angle and add the three Hydrogen in direction of the bond (height of the tetraoid)
343 FirstOtherAtom = new atom();
344 SecondOtherAtom = new atom();
345 ThirdOtherAtom = new atom();
346 FirstOtherAtom->type = elemente->FindElement(1);
347 SecondOtherAtom->type = elemente->FindElement(1);
348 ThirdOtherAtom->type = elemente->FindElement(1);
349 FirstOtherAtom->v.CopyVector(&TopReplacement->v); // copy velocity
350 FirstOtherAtom->FixedIon = TopReplacement->FixedIon;
351 SecondOtherAtom->v.CopyVector(&TopReplacement->v); // copy velocity
352 SecondOtherAtom->FixedIon = TopReplacement->FixedIon;
353 ThirdOtherAtom->v.CopyVector(&TopReplacement->v); // copy velocity
354 ThirdOtherAtom->FixedIon = TopReplacement->FixedIon;
355 FirstOtherAtom->father = NULL; // we are just an added hydrogen with no father
356 SecondOtherAtom->father = NULL; // we are just an added hydrogen with no father
357 ThirdOtherAtom->father = NULL; // we are just an added hydrogen with no father
358
359 // we need to vectors orthonormal the InBondvector
360 AllWentWell = AllWentWell && Orthovector1.GetOneNormalVector(&InBondvector);
361// *out << Verbose(3) << "Orthovector1: ";
362// Orthovector1.Output(out);
363// *out << endl;
364 AllWentWell = AllWentWell && Orthovector2.MakeNormalVector(&InBondvector, &Orthovector1);
365// *out << Verbose(3) << "Orthovector2: ";
366// Orthovector2.Output(out);
367// *out << endl;
368
369 // create correct coordination for the three atoms
370 alpha = (TopOrigin->type->HBondAngle[2])/180.*M_PI/2.; // retrieve triple bond angle from database
371 l = BondRescale; // desired bond length
372 b = 2.*l*sin(alpha); // base length of isosceles triangle
373 d = l*sqrt(cos(alpha)*cos(alpha) - sin(alpha)*sin(alpha)/3.); // length for InBondvector
374 f = b/sqrt(3.); // length for Orthvector1
375 g = b/2.; // length for Orthvector2
376// *out << Verbose(3) << "Bond length and half-angle: " << l << ", " << alpha << "\t (b,d,f,g) = " << b << ", " << d << ", " << f << ", " << g << ", " << endl;
377// *out << Verbose(3) << "The three Bond lengths: " << sqrt(d*d+f*f) << ", " << sqrt(d*d+(-0.5*f)*(-0.5*f)+g*g) << ", " << sqrt(d*d+(-0.5*f)*(-0.5*f)+g*g) << endl;
378 factors[0] = d;
379 factors[1] = f;
380 factors[2] = 0.;
381 FirstOtherAtom->x.LinearCombinationOfVectors(&InBondvector, &Orthovector1, &Orthovector2, factors);
382 factors[1] = -0.5*f;
383 factors[2] = g;
384 SecondOtherAtom->x.LinearCombinationOfVectors(&InBondvector, &Orthovector1, &Orthovector2, factors);
385 factors[2] = -g;
386 ThirdOtherAtom->x.LinearCombinationOfVectors(&InBondvector, &Orthovector1, &Orthovector2, factors);
387
388 // rescale each to correct BondDistance
389// FirstOtherAtom->x.Scale(&BondRescale);
390// SecondOtherAtom->x.Scale(&BondRescale);
391// ThirdOtherAtom->x.Scale(&BondRescale);
392
393 // and relative to *origin atom
394 FirstOtherAtom->x.AddVector(&TopOrigin->x);
395 SecondOtherAtom->x.AddVector(&TopOrigin->x);
396 ThirdOtherAtom->x.AddVector(&TopOrigin->x);
397
398 // ... and add to molecule
399 AllWentWell = AllWentWell && AddAtom(FirstOtherAtom);
400 AllWentWell = AllWentWell && AddAtom(SecondOtherAtom);
401 AllWentWell = AllWentWell && AddAtom(ThirdOtherAtom);
402// *out << Verbose(4) << "Added " << *FirstOtherAtom << " at: ";
403// FirstOtherAtom->x.Output(out);
404// *out << endl;
405// *out << Verbose(4) << "Added " << *SecondOtherAtom << " at: ";
406// SecondOtherAtom->x.Output(out);
407// *out << endl;
408// *out << Verbose(4) << "Added " << *ThirdOtherAtom << " at: ";
409// ThirdOtherAtom->x.Output(out);
410// *out << endl;
411 Binder = AddBond(BottomOrigin, FirstOtherAtom, 1);
412 Binder->Cyclic = false;
413 Binder->Type = TreeEdge;
414 Binder = AddBond(BottomOrigin, SecondOtherAtom, 1);
415 Binder->Cyclic = false;
416 Binder->Type = TreeEdge;
417 Binder = AddBond(BottomOrigin, ThirdOtherAtom, 1);
418 Binder->Cyclic = false;
419 Binder->Type = TreeEdge;
420 break;
421 default:
422 cerr << "ERROR: BondDegree does not state single, double or triple bond!" << endl;
423 AllWentWell = false;
424 break;
425 }
426
427// *out << Verbose(3) << "End of AddHydrogenReplacementAtom." << endl;
428 return AllWentWell;
429};
430
431/** Adds given atom \a *pointer from molecule list.
432 * Increases molecule::last_atom and gives last number to added atom.
433 * \param filename name and path of xyz file
434 * \return true - succeeded, false - file not found
435 */
436bool molecule::AddXYZFile(string filename)
437{
438 istringstream *input = NULL;
439 int NumberOfAtoms = 0; // atom number in xyz read
440 int i, j; // loop variables
441 atom *Walker = NULL; // pointer to added atom
442 char shorthand[3]; // shorthand for atom name
443 ifstream xyzfile; // xyz file
444 string line; // currently parsed line
445 double x[3]; // atom coordinates
446
447 xyzfile.open(filename.c_str());
448 if (!xyzfile)
449 return false;
450
451 getline(xyzfile,line,'\n'); // Read numer of atoms in file
452 input = new istringstream(line);
453 *input >> NumberOfAtoms;
454 cout << Verbose(0) << "Parsing " << NumberOfAtoms << " atoms in file." << endl;
455 getline(xyzfile,line,'\n'); // Read comment
456 cout << Verbose(1) << "Comment: " << line << endl;
457
458 if (MDSteps == 0) // no atoms yet present
459 MDSteps++;
460 for(i=0;i<NumberOfAtoms;i++){
461 Walker = new atom;
462 getline(xyzfile,line,'\n');
463 istringstream *item = new istringstream(line);
464 //istringstream input(line);
465 //cout << Verbose(1) << "Reading: " << line << endl;
466 *item >> shorthand;
467 *item >> x[0];
468 *item >> x[1];
469 *item >> x[2];
470 Walker->type = elemente->FindElement(shorthand);
471 if (Walker->type == NULL) {
472 cerr << "Could not parse the element at line: '" << line << "', setting to H.";
473 Walker->type = elemente->FindElement(1);
474 }
475 if (Trajectories[Walker].R.size() <= (unsigned int)MDSteps) {
476 Trajectories[Walker].R.resize(MDSteps+10);
477 Trajectories[Walker].U.resize(MDSteps+10);
478 Trajectories[Walker].F.resize(MDSteps+10);
479 }
480 for(j=NDIM;j--;) {
481 Walker->x.x[j] = x[j];
482 Trajectories[Walker].R.at(MDSteps-1).x[j] = x[j];
483 Trajectories[Walker].U.at(MDSteps-1).x[j] = 0;
484 Trajectories[Walker].F.at(MDSteps-1).x[j] = 0;
485 }
486 AddAtom(Walker); // add to molecule
487 delete(item);
488 }
489 xyzfile.close();
490 delete(input);
491 return true;
492};
493
494/** Creates a copy of this molecule.
495 * \return copy of molecule
496 */
497molecule *molecule::CopyMolecule()
498{
499 molecule *copy = new molecule(elemente);
500 atom *CurrentAtom = NULL;
501 atom *LeftAtom = NULL, *RightAtom = NULL;
502 atom *Walker = NULL;
503
504 // copy all atoms
505 Walker = start;
506 while(Walker->next != end) {
507 Walker = Walker->next;
508 CurrentAtom = copy->AddCopyAtom(Walker);
509 }
510
511 // copy all bonds
512 bond *Binder = first;
513 bond *NewBond = NULL;
514 while(Binder->next != last) {
515 Binder = Binder->next;
516 // get the pendant atoms of current bond in the copy molecule
517 LeftAtom = copy->start;
518 while (LeftAtom->next != copy->end) {
519 LeftAtom = LeftAtom->next;
520 if (LeftAtom->father == Binder->leftatom)
521 break;
522 }
523 RightAtom = copy->start;
524 while (RightAtom->next != copy->end) {
525 RightAtom = RightAtom->next;
526 if (RightAtom->father == Binder->rightatom)
527 break;
528 }
529 NewBond = copy->AddBond(LeftAtom, RightAtom, Binder->BondDegree);
530 NewBond->Cyclic = Binder->Cyclic;
531 if (Binder->Cyclic)
532 copy->NoCyclicBonds++;
533 NewBond->Type = Binder->Type;
534 }
535 // correct fathers
536 Walker = copy->start;
537 while(Walker->next != copy->end) {
538 Walker = Walker->next;
539 if (Walker->father->father == Walker->father) // same atom in copy's father points to itself
540 Walker->father = Walker; // set father to itself (copy of a whole molecule)
541 else
542 Walker->father = Walker->father->father; // set father to original's father
543 }
544 // copy values
545 copy->CountAtoms((ofstream *)&cout);
546 copy->CountElements();
547 if (first->next != last) { // if adjaceny list is present
548 copy->BondDistance = BondDistance;
549 copy->CreateListOfBondsPerAtom((ofstream *)&cout);
550 }
551
552 return copy;
553};
554
555/** Adds a bond to a the molecule specified by two atoms, \a *first and \a *second.
556 * Also updates molecule::BondCount and molecule::NoNonBonds.
557 * \param *first first atom in bond
558 * \param *second atom in bond
559 * \return pointer to bond or NULL on failure
560 */
561bond * molecule::AddBond(atom *atom1, atom *atom2, int degree=1)
562{
563 bond *Binder = NULL;
564 if ((atom1 != NULL) && (FindAtom(atom1->nr) != NULL) && (atom2 != NULL) && (FindAtom(atom2->nr) != NULL)) {
565 Binder = new bond(atom1, atom2, degree, BondCount++);
566 if ((atom1->type != NULL) && (atom1->type->Z != 1) && (atom2->type != NULL) && (atom2->type->Z != 1))
567 NoNonBonds++;
568 add(Binder, last);
569 } else {
570 cerr << Verbose(1) << "ERROR: Could not add bond between " << atom1->Name << " and " << atom2->Name << " as one or both are not present in the molecule." << endl;
571 }
572 return Binder;
573};
574
575/** Remove bond from bond chain list.
576 * \todo Function not implemented yet
577 * \param *pointer bond pointer
578 * \return true - bound found and removed, false - bond not found/removed
579 */
580bool molecule::RemoveBond(bond *pointer)
581{
582 //cerr << Verbose(1) << "molecule::RemoveBond: Function not implemented yet." << endl;
583 removewithoutcheck(pointer);
584 return true;
585};
586
587/** Remove every bond from bond chain list that atom \a *BondPartner is a constituent of.
588 * \todo Function not implemented yet
589 * \param *BondPartner atom to be removed
590 * \return true - bounds found and removed, false - bonds not found/removed
591 */
592bool molecule::RemoveBonds(atom *BondPartner)
593{
594 cerr << Verbose(1) << "molecule::RemoveBond: Function not implemented yet." << endl;
595 return false;
596};
597
598/** Sets the molecule::cell_size to the components of \a *dim (rectangular box)
599 * \param *dim vector class
600 */
601void molecule::SetBoxDimension(Vector *dim)
602{
603 cell_size[0] = dim->x[0];
604 cell_size[1] = 0.;
605 cell_size[2] = dim->x[1];
606 cell_size[3] = 0.;
607 cell_size[4] = 0.;
608 cell_size[5] = dim->x[2];
609};
610
611/** Centers the molecule in the box whose lengths are defined by vector \a *BoxLengths.
612 * \param *out output stream for debugging
613 * \param *BoxLengths box lengths
614 */
615bool molecule::CenterInBox(ofstream *out, Vector *BoxLengths)
616{
617 bool status = true;
618 atom *ptr = NULL;
619 Vector *min = new Vector;
620 Vector *max = new Vector;
621
622 // gather min and max for each axis
623 ptr = start->next; // start at first in list
624 if (ptr != end) { //list not empty?
625 for (int i=NDIM;i--;) {
626 max->x[i] = ptr->x.x[i];
627 min->x[i] = ptr->x.x[i];
628 }
629 while (ptr->next != end) { // continue with second if present
630 ptr = ptr->next;
631 //ptr->Output(1,1,out);
632 for (int i=NDIM;i--;) {
633 max->x[i] = (max->x[i] < ptr->x.x[i]) ? ptr->x.x[i] : max->x[i];
634 min->x[i] = (min->x[i] > ptr->x.x[i]) ? ptr->x.x[i] : min->x[i];
635 }
636 }
637 }
638 // sanity check
639 for(int i=NDIM;i--;) {
640 if (max->x[i] - min->x[i] > BoxLengths->x[i])
641 status = false;
642 }
643 // warn if check failed
644 if (!status)
645 *out << "WARNING: molecule is bigger than defined box!" << endl;
646 else { // else center in box
647 max->AddVector(min);
648 max->Scale(-1.);
649 max->AddVector(BoxLengths);
650 max->Scale(0.5);
651 Translate(max);
652 }
653
654 // free and exit
655 delete(min);
656 delete(max);
657 return status;
658};
659
660/** Centers the edge of the atoms at (0,0,0).
661 * \param *out output stream for debugging
662 * \param *max coordinates of other edge, specifying box dimensions.
663 */
664void molecule::CenterEdge(ofstream *out, Vector *max)
665{
666 Vector *min = new Vector;
667
668// *out << Verbose(3) << "Begin of CenterEdge." << endl;
669 atom *ptr = start->next; // start at first in list
670 if (ptr != end) { //list not empty?
671 for (int i=NDIM;i--;) {
672 max->x[i] = ptr->x.x[i];
673 min->x[i] = ptr->x.x[i];
674 }
675 while (ptr->next != end) { // continue with second if present
676 ptr = ptr->next;
677 //ptr->Output(1,1,out);
678 for (int i=NDIM;i--;) {
679 max->x[i] = (max->x[i] < ptr->x.x[i]) ? ptr->x.x[i] : max->x[i];
680 min->x[i] = (min->x[i] > ptr->x.x[i]) ? ptr->x.x[i] : min->x[i];
681 }
682 }
683// *out << Verbose(4) << "Maximum is ";
684// max->Output(out);
685// *out << ", Minimum is ";
686// min->Output(out);
687// *out << endl;
688 min->Scale(-1.);
689 max->AddVector(min);
690 Translate(min);
691 }
692 delete(min);
693// *out << Verbose(3) << "End of CenterEdge." << endl;
694};
695
696/** Centers the center of the atoms at (0,0,0).
697 * \param *out output stream for debugging
698 * \param *center return vector for translation vector
699 */
700void molecule::CenterOrigin(ofstream *out, Vector *center)
701{
702 int Num = 0;
703 atom *ptr = start->next; // start at first in list
704
705 for(int i=NDIM;i--;) // zero center vector
706 center->x[i] = 0.;
707
708 if (ptr != end) { //list not empty?
709 while (ptr->next != end) { // continue with second if present
710 ptr = ptr->next;
711 Num++;
712 center->AddVector(&ptr->x);
713 }
714 center->Scale(-1./Num); // divide through total number (and sign for direction)
715 Translate(center);
716 }
717};
718
719/** Returns vector pointing to center of gravity.
720 * \param *out output stream for debugging
721 * \return pointer to center of gravity vector
722 */
723Vector * molecule::DetermineCenterOfAll(ofstream *out)
724{
725 atom *ptr = start->next; // start at first in list
726 Vector *a = new Vector();
727 Vector tmp;
728 double Num = 0;
729
730 a->Zero();
731
732 if (ptr != end) { //list not empty?
733 while (ptr->next != end) { // continue with second if present
734 ptr = ptr->next;
735 Num += 1.;
736 tmp.CopyVector(&ptr->x);
737 a->AddVector(&tmp);
738 }
739 a->Scale(-1./Num); // divide through total mass (and sign for direction)
740 }
741 //cout << Verbose(1) << "Resulting center of gravity: ";
742 //a->Output(out);
743 //cout << endl;
744 return a;
745};
746
747/** Returns vector pointing to center of gravity.
748 * \param *out output stream for debugging
749 * \return pointer to center of gravity vector
750 */
751Vector * molecule::DetermineCenterOfGravity(ofstream *out)
752{
753 atom *ptr = start->next; // start at first in list
754 Vector *a = new Vector();
755 Vector tmp;
756 double Num = 0;
757
758 a->Zero();
759
760 if (ptr != end) { //list not empty?
761 while (ptr->next != end) { // continue with second if present
762 ptr = ptr->next;
763 Num += ptr->type->mass;
764 tmp.CopyVector(&ptr->x);
765 tmp.Scale(ptr->type->mass); // scale by mass
766 a->AddVector(&tmp);
767 }
768 a->Scale(-1./Num); // divide through total mass (and sign for direction)
769 }
770// *out << Verbose(1) << "Resulting center of gravity: ";
771// a->Output(out);
772// *out << endl;
773 return a;
774};
775
776/** Centers the center of gravity of the atoms at (0,0,0).
777 * \param *out output stream for debugging
778 * \param *center return vector for translation vector
779 */
780void molecule::CenterGravity(ofstream *out, Vector *center)
781{
782 if (center == NULL) {
783 DetermineCenter(*center);
784 Translate(center);
785 delete(center);
786 } else {
787 Translate(center);
788 }
789};
790
791/** Scales all atoms by \a *factor.
792 * \param *factor pointer to scaling factor
793 */
794void molecule::Scale(double **factor)
795{
796 atom *ptr = start;
797
798 while (ptr->next != end) {
799 ptr = ptr->next;
800 for (int j=0;j<MDSteps;j++)
801 Trajectories[ptr].R.at(j).Scale(factor);
802 ptr->x.Scale(factor);
803 }
804};
805
806/** Translate all atoms by given vector.
807 * \param trans[] translation vector.
808 */
809void molecule::Translate(const Vector *trans)
810{
811 atom *ptr = start;
812
813 while (ptr->next != end) {
814 ptr = ptr->next;
815 for (int j=0;j<MDSteps;j++)
816 Trajectories[ptr].R.at(j).Translate(trans);
817 ptr->x.Translate(trans);
818 }
819};
820
821/** Mirrors all atoms against a given plane.
822 * \param n[] normal vector of mirror plane.
823 */
824void molecule::Mirror(const Vector *n)
825{
826 atom *ptr = start;
827
828 while (ptr->next != end) {
829 ptr = ptr->next;
830 for (int j=0;j<MDSteps;j++)
831 Trajectories[ptr].R.at(j).Mirror(n);
832 ptr->x.Mirror(n);
833 }
834};
835
836/** Determines center of molecule (yet not considering atom masses).
837 * \param Center reference to return vector
838 */
839void molecule::DetermineCenter(Vector &Center)
840{
841 atom *Walker = start;
842 bond *Binder = NULL;
843 double *matrix = ReturnFullMatrixforSymmetric(cell_size);
844 double tmp;
845 bool flag;
846 Vector Testvector, Translationvector;
847
848 do {
849 Center.Zero();
850 flag = true;
851 while (Walker->next != end) {
852 Walker = Walker->next;
853#ifdef ADDHYDROGEN
854 if (Walker->type->Z != 1) {
855#endif
856 Testvector.CopyVector(&Walker->x);
857 Testvector.InverseMatrixMultiplication(matrix);
858 Translationvector.Zero();
859 for (int i=0;i<NumberOfBondsPerAtom[Walker->nr]; i++) {
860 Binder = ListOfBondsPerAtom[Walker->nr][i];
861 if (Walker->nr < Binder->GetOtherAtom(Walker)->nr) // otherwise we shift one to, the other fro and gain nothing
862 for (int j=0;j<NDIM;j++) {
863 tmp = Walker->x.x[j] - Binder->GetOtherAtom(Walker)->x.x[j];
864 if ((fabs(tmp)) > BondDistance) {
865 flag = false;
866 cout << Verbose(0) << "Hit: atom " << Walker->Name << " in bond " << *Binder << " has to be shifted due to " << tmp << "." << endl;
867 if (tmp > 0)
868 Translationvector.x[j] -= 1.;
869 else
870 Translationvector.x[j] += 1.;
871 }
872 }
873 }
874 Testvector.AddVector(&Translationvector);
875 Testvector.MatrixMultiplication(matrix);
876 Center.AddVector(&Testvector);
877 cout << Verbose(1) << "vector is: ";
878 Testvector.Output((ofstream *)&cout);
879 cout << endl;
880#ifdef ADDHYDROGEN
881 // now also change all hydrogens
882 for (int i=0;i<NumberOfBondsPerAtom[Walker->nr]; i++) {
883 Binder = ListOfBondsPerAtom[Walker->nr][i];
884 if (Binder->GetOtherAtom(Walker)->type->Z == 1) {
885 Testvector.CopyVector(&Binder->GetOtherAtom(Walker)->x);
886 Testvector.InverseMatrixMultiplication(matrix);
887 Testvector.AddVector(&Translationvector);
888 Testvector.MatrixMultiplication(matrix);
889 Center.AddVector(&Testvector);
890 cout << Verbose(1) << "Hydrogen vector is: ";
891 Testvector.Output((ofstream *)&cout);
892 cout << endl;
893 }
894 }
895 }
896#endif
897 }
898 } while (!flag);
899 Free((void **)&matrix, "molecule::DetermineCenter: *matrix");
900 Center.Scale(1./(double)AtomCount);
901};
902
903/** Transforms/Rotates the given molecule into its principal axis system.
904 * \param *out output stream for debugging
905 * \param DoRotate whether to rotate (true) or only to determine the PAS.
906 */
907void molecule::PrincipalAxisSystem(ofstream *out, bool DoRotate)
908{
909 atom *ptr = start; // start at first in list
910 double InertiaTensor[NDIM*NDIM];
911 Vector *CenterOfGravity = DetermineCenterOfGravity(out);
912
913 CenterGravity(out, CenterOfGravity);
914
915 // reset inertia tensor
916 for(int i=0;i<NDIM*NDIM;i++)
917 InertiaTensor[i] = 0.;
918
919 // sum up inertia tensor
920 while (ptr->next != end) {
921 ptr = ptr->next;
922 Vector x;
923 x.CopyVector(&ptr->x);
924 //x.SubtractVector(CenterOfGravity);
925 InertiaTensor[0] += ptr->type->mass*(x.x[1]*x.x[1] + x.x[2]*x.x[2]);
926 InertiaTensor[1] += ptr->type->mass*(-x.x[0]*x.x[1]);
927 InertiaTensor[2] += ptr->type->mass*(-x.x[0]*x.x[2]);
928 InertiaTensor[3] += ptr->type->mass*(-x.x[1]*x.x[0]);
929 InertiaTensor[4] += ptr->type->mass*(x.x[0]*x.x[0] + x.x[2]*x.x[2]);
930 InertiaTensor[5] += ptr->type->mass*(-x.x[1]*x.x[2]);
931 InertiaTensor[6] += ptr->type->mass*(-x.x[2]*x.x[0]);
932 InertiaTensor[7] += ptr->type->mass*(-x.x[2]*x.x[1]);
933 InertiaTensor[8] += ptr->type->mass*(x.x[0]*x.x[0] + x.x[1]*x.x[1]);
934 }
935 // print InertiaTensor for debugging
936 *out << "The inertia tensor is:" << endl;
937 for(int i=0;i<NDIM;i++) {
938 for(int j=0;j<NDIM;j++)
939 *out << InertiaTensor[i*NDIM+j] << " ";
940 *out << endl;
941 }
942 *out << endl;
943
944 // diagonalize to determine principal axis system
945 gsl_eigen_symmv_workspace *T = gsl_eigen_symmv_alloc(NDIM);
946 gsl_matrix_view m = gsl_matrix_view_array(InertiaTensor, NDIM, NDIM);
947 gsl_vector *eval = gsl_vector_alloc(NDIM);
948 gsl_matrix *evec = gsl_matrix_alloc(NDIM, NDIM);
949 gsl_eigen_symmv(&m.matrix, eval, evec, T);
950 gsl_eigen_symmv_free(T);
951 gsl_eigen_symmv_sort(eval, evec, GSL_EIGEN_SORT_ABS_DESC);
952
953 for(int i=0;i<NDIM;i++) {
954 *out << Verbose(1) << "eigenvalue = " << gsl_vector_get(eval, i);
955 *out << ", eigenvector = (" << evec->data[i * evec->tda + 0] << "," << evec->data[i * evec->tda + 1] << "," << evec->data[i * evec->tda + 2] << ")" << endl;
956 }
957
958 // check whether we rotate or not
959 if (DoRotate) {
960 *out << Verbose(1) << "Transforming molecule into PAS ... ";
961 // the eigenvectors specify the transformation matrix
962 ptr = start;
963 while (ptr->next != end) {
964 ptr = ptr->next;
965 for (int j=0;j<MDSteps;j++)
966 Trajectories[ptr].R.at(j).MatrixMultiplication(evec->data);
967 ptr->x.MatrixMultiplication(evec->data);
968 }
969 *out << "done." << endl;
970
971 // summing anew for debugging (resulting matrix has to be diagonal!)
972 // reset inertia tensor
973 for(int i=0;i<NDIM*NDIM;i++)
974 InertiaTensor[i] = 0.;
975
976 // sum up inertia tensor
977 ptr = start;
978 while (ptr->next != end) {
979 ptr = ptr->next;
980 Vector x;
981 x.CopyVector(&ptr->x);
982 //x.SubtractVector(CenterOfGravity);
983 InertiaTensor[0] += ptr->type->mass*(x.x[1]*x.x[1] + x.x[2]*x.x[2]);
984 InertiaTensor[1] += ptr->type->mass*(-x.x[0]*x.x[1]);
985 InertiaTensor[2] += ptr->type->mass*(-x.x[0]*x.x[2]);
986 InertiaTensor[3] += ptr->type->mass*(-x.x[1]*x.x[0]);
987 InertiaTensor[4] += ptr->type->mass*(x.x[0]*x.x[0] + x.x[2]*x.x[2]);
988 InertiaTensor[5] += ptr->type->mass*(-x.x[1]*x.x[2]);
989 InertiaTensor[6] += ptr->type->mass*(-x.x[2]*x.x[0]);
990 InertiaTensor[7] += ptr->type->mass*(-x.x[2]*x.x[1]);
991 InertiaTensor[8] += ptr->type->mass*(x.x[0]*x.x[0] + x.x[1]*x.x[1]);
992 }
993 // print InertiaTensor for debugging
994 *out << "The inertia tensor is:" << endl;
995 for(int i=0;i<NDIM;i++) {
996 for(int j=0;j<NDIM;j++)
997 *out << InertiaTensor[i*NDIM+j] << " ";
998 *out << endl;
999 }
1000 *out << endl;
1001 }
1002
1003 // free everything
1004 delete(CenterOfGravity);
1005 gsl_vector_free(eval);
1006 gsl_matrix_free(evec);
1007};
1008
1009/** Evaluates the potential energy used for constrained molecular dynamics.
1010 * \f$V_i^{con} = c^{bond} \cdot | r_{P(i)} - R_i | + sum_{i \neq j} C^{min} \cdot \frac{1}{C_{ij}} + C^{inj} \Bigl (1 - \theta \bigl (\prod_{i \neq j} (P(i) - P(j)) \bigr ) \Bigr )\f$
1011 * where the first term points to the target in minimum distance, the second is a penalty for trajectories lying too close to each other (\f$C_{ij}$ is minimum distance between
1012 * trajectories i and j) and the third term is a penalty for two atoms trying to each the same target point.
1013 * Note that for the second term we have to solve the following linear system:
1014 * \f$-c_1 \cdot n_1 + c_2 \cdot n_2 + C \cdot n_3 = - p_2 + p_1\f$, where \f$c_1\f$, \f$c_2\f$ and \f$C\f$ are constants,
1015 * offset vector \f$p_1\f$ in direction \f$n_1\f$, offset vector \f$p_2\f$ in direction \f$n_2\f$,
1016 * \f$n_3\f$ is the normal vector to both directions. \f$C\f$ would be the minimum distance between the two lines.
1017 * \sa molecule::MinimiseConstrainedPotential(), molecule::VerletForceIntegration()
1018 * \param *out output stream for debugging
1019 * \param *PermutationMap gives target ptr for each atom, array of size molecule::AtomCount (this is "x" in \f$ V^{con}(x) \f$ )
1020 * \param startstep start configuration (MDStep in molecule::trajectories)
1021 * \param endstep end configuration (MDStep in molecule::trajectories)
1022 * \param *constants constant in front of each term
1023 * \param IsAngstroem whether coordinates are in angstroem (true) or bohrradius (false)
1024 * \return potential energy
1025 * \note This routine is scaling quadratically which is not optimal.
1026 * \todo There's a bit double counting going on for the first time, bu nothing to worry really about.
1027 */
1028double molecule::ConstrainedPotential(ofstream *out, atom **PermutationMap, int startstep, int endstep, double *constants, bool IsAngstroem)
1029{
1030 double result = 0., tmp, Norm1, Norm2;
1031 atom *Walker = NULL, *Runner = NULL, *Sprinter = NULL;
1032 Vector trajectory1, trajectory2, normal, TestVector;
1033 gsl_matrix *A = gsl_matrix_alloc(NDIM,NDIM);
1034 gsl_vector *x = gsl_vector_alloc(NDIM);
1035
1036 // go through every atom
1037 Walker = start;
1038 while (Walker->next != end) {
1039 Walker = Walker->next;
1040 // first term: distance to target
1041 Runner = PermutationMap[Walker->nr]; // find target point
1042 tmp = (Trajectories[Walker].R.at(startstep).Distance(&Trajectories[Runner].R.at(endstep)));
1043 tmp *= IsAngstroem ? 1. : 1./AtomicLengthToAngstroem;
1044 result += constants[0] * tmp;
1045 //*out << Verbose(4) << "Adding " << tmp*constants[0] << "." << endl;
1046
1047 // second term: sum of distances to other trajectories
1048 Runner = start;
1049 while (Runner->next != end) {
1050 Runner = Runner->next;
1051 if (Runner == Walker) // hence, we only go up to the Walker, not beyond (similar to i=0; i<j; i++)
1052 break;
1053 // determine normalized trajectories direction vector (n1, n2)
1054 Sprinter = PermutationMap[Walker->nr]; // find first target point
1055 trajectory1.CopyVector(&Trajectories[Sprinter].R.at(endstep));
1056 trajectory1.SubtractVector(&Trajectories[Walker].R.at(startstep));
1057 trajectory1.Normalize();
1058 Norm1 = trajectory1.Norm();
1059 Sprinter = PermutationMap[Runner->nr]; // find second target point
1060 trajectory2.CopyVector(&Trajectories[Sprinter].R.at(endstep));
1061 trajectory2.SubtractVector(&Trajectories[Runner].R.at(startstep));
1062 trajectory2.Normalize();
1063 Norm2 = trajectory1.Norm();
1064 // check whether either is zero()
1065 if ((Norm1 < MYEPSILON) && (Norm2 < MYEPSILON)) {
1066 tmp = Trajectories[Walker].R.at(startstep).Distance(&Trajectories[Runner].R.at(startstep));
1067 } else if (Norm1 < MYEPSILON) {
1068 Sprinter = PermutationMap[Walker->nr]; // find first target point
1069 trajectory1.CopyVector(&Trajectories[Sprinter].R.at(endstep)); // copy first offset
1070 trajectory1.SubtractVector(&Trajectories[Runner].R.at(startstep)); // subtract second offset
1071 trajectory2.Scale( trajectory1.ScalarProduct(&trajectory2) ); // trajectory2 is scaled to unity, hence we don't need to divide by anything
1072 trajectory1.SubtractVector(&trajectory2); // project the part in norm direction away
1073 tmp = trajectory1.Norm(); // remaining norm is distance
1074 } else if (Norm2 < MYEPSILON) {
1075 Sprinter = PermutationMap[Runner->nr]; // find second target point
1076 trajectory2.CopyVector(&Trajectories[Sprinter].R.at(endstep)); // copy second offset
1077 trajectory2.SubtractVector(&Trajectories[Walker].R.at(startstep)); // subtract first offset
1078 trajectory1.Scale( trajectory2.ScalarProduct(&trajectory1) ); // trajectory1 is scaled to unity, hence we don't need to divide by anything
1079 trajectory2.SubtractVector(&trajectory1); // project the part in norm direction away
1080 tmp = trajectory2.Norm(); // remaining norm is distance
1081 } else if ((fabs(trajectory1.ScalarProduct(&trajectory2)/Norm1/Norm2) - 1.) < MYEPSILON) { // check whether they're linear dependent
1082// *out << Verbose(3) << "Both trajectories of " << *Walker << " and " << *Runner << " are linear dependent: ";
1083// *out << trajectory1;
1084// *out << " and ";
1085// *out << trajectory2;
1086 tmp = Trajectories[Walker].R.at(startstep).Distance(&Trajectories[Runner].R.at(startstep));
1087// *out << " with distance " << tmp << "." << endl;
1088 } else { // determine distance by finding minimum distance
1089// *out << Verbose(3) << "Both trajectories of " << *Walker << " and " << *Runner << " are linear independent ";
1090// *out << endl;
1091// *out << "First Trajectory: ";
1092// *out << trajectory1 << endl;
1093// *out << "Second Trajectory: ";
1094// *out << trajectory2 << endl;
1095 // determine normal vector for both
1096 normal.MakeNormalVector(&trajectory1, &trajectory2);
1097 // print all vectors for debugging
1098// *out << "Normal vector in between: ";
1099// *out << normal << endl;
1100 // setup matrix
1101 for (int i=NDIM;i--;) {
1102 gsl_matrix_set(A, 0, i, trajectory1.x[i]);
1103 gsl_matrix_set(A, 1, i, trajectory2.x[i]);
1104 gsl_matrix_set(A, 2, i, normal.x[i]);
1105 gsl_vector_set(x,i, (Trajectories[Walker].R.at(startstep).x[i] - Trajectories[Runner].R.at(startstep).x[i]));
1106 }
1107 // solve the linear system by Householder transformations
1108 gsl_linalg_HH_svx(A, x);
1109 // distance from last component
1110 tmp = gsl_vector_get(x,2);
1111// *out << " with distance " << tmp << "." << endl;
1112 // test whether we really have the intersection (by checking on c_1 and c_2)
1113 TestVector.CopyVector(&Trajectories[Runner].R.at(startstep));
1114 trajectory2.Scale(gsl_vector_get(x,1));
1115 TestVector.AddVector(&trajectory2);
1116 normal.Scale(gsl_vector_get(x,2));
1117 TestVector.AddVector(&normal);
1118 TestVector.SubtractVector(&Trajectories[Walker].R.at(startstep));
1119 trajectory1.Scale(gsl_vector_get(x,0));
1120 TestVector.SubtractVector(&trajectory1);
1121 if (TestVector.Norm() < MYEPSILON) {
1122// *out << Verbose(2) << "Test: ok.\tDistance of " << tmp << " is correct." << endl;
1123 } else {
1124// *out << Verbose(2) << "Test: failed.\tIntersection is off by ";
1125// *out << TestVector;
1126// *out << "." << endl;
1127 }
1128 }
1129 // add up
1130 tmp *= IsAngstroem ? 1. : 1./AtomicLengthToAngstroem;
1131 if (fabs(tmp) > MYEPSILON) {
1132 result += constants[1] * 1./tmp;
1133 //*out << Verbose(4) << "Adding " << 1./tmp*constants[1] << "." << endl;
1134 }
1135 }
1136
1137 // third term: penalty for equal targets
1138 Runner = start;
1139 while (Runner->next != end) {
1140 Runner = Runner->next;
1141 if ((PermutationMap[Walker->nr] == PermutationMap[Runner->nr]) && (Walker->nr < Runner->nr)) {
1142 Sprinter = PermutationMap[Walker->nr];
1143// *out << *Walker << " and " << *Runner << " are heading to the same target at ";
1144// *out << Trajectories[Sprinter].R.at(endstep);
1145// *out << ", penalting." << endl;
1146 result += constants[2];
1147 //*out << Verbose(4) << "Adding " << constants[2] << "." << endl;
1148 }
1149 }
1150 }
1151
1152 return result;
1153};
1154
1155void PrintPermutationMap(ofstream *out, atom **PermutationMap, int Nr)
1156{
1157 stringstream zeile1, zeile2;
1158 int *DoubleList = (int *) Malloc(Nr*sizeof(int), "PrintPermutationMap: *DoubleList");
1159 int doubles = 0;
1160 for (int i=0;i<Nr;i++)
1161 DoubleList[i] = 0;
1162 zeile1 << "PermutationMap: ";
1163 zeile2 << " ";
1164 for (int i=0;i<Nr;i++) {
1165 DoubleList[PermutationMap[i]->nr]++;
1166 zeile1 << i << " ";
1167 zeile2 << PermutationMap[i]->nr << " ";
1168 }
1169 for (int i=0;i<Nr;i++)
1170 if (DoubleList[i] > 1)
1171 doubles++;
1172 // *out << "Found " << doubles << " Doubles." << endl;
1173 Free((void **)&DoubleList, "PrintPermutationMap: *DoubleList");
1174// *out << zeile1.str() << endl << zeile2.str() << endl;
1175};
1176
1177/** Minimises the extra potential for constrained molecular dynamics and gives forces and the constrained potential energy.
1178 * We do the following:
1179 * -# Generate a distance list from all source to all target points
1180 * -# Sort this per source point
1181 * -# Take for each source point the target point with minimum distance, use this as initial permutation
1182 * -# check whether molecule::ConstrainedPotential() is greater than injective penalty
1183 * -# If so, we go through each source point, stepping down in the sorted target point distance list and re-checking potential.
1184 * -# Next, we only apply transformations that keep the injectivity of the permutations list.
1185 * -# Hence, for one source point we step down the ladder and seek the corresponding owner of this new target
1186 * point and try to change it for one with lesser distance, or for the next one with greater distance, but only
1187 * if this decreases the conditional potential.
1188 * -# finished.
1189 * -# Then, we calculate the forces by taking the spatial derivative, where we scale the potential to such a degree,
1190 * that the total force is always pointing in direction of the constraint force (ensuring that we move in the
1191 * right direction).
1192 * -# Finally, we calculate the potential energy and return.
1193 * \param *out output stream for debugging
1194 * \param **PermutationMap on return: mapping between the atom label of the initial and the final configuration
1195 * \param startstep current MD step giving initial position between which and \a endstep we perform the constrained MD (as further steps are always concatenated)
1196 * \param endstep step giving final position in constrained MD
1197 * \param IsAngstroem whether coordinates are in angstroem (true) or bohrradius (false)
1198 * \sa molecule::VerletForceIntegration()
1199 * \return potential energy (and allocated **PermutationMap (array of molecule::AtomCount ^2)
1200 * \todo The constrained potential's constants are set to fixed values right now, but they should scale based on checks of the system in order
1201 * to ensure they're properties (e.g. constants[2] always greater than the energy of the system).
1202 * \bug this all is not O(N log N) but O(N^2)
1203 */
1204double molecule::MinimiseConstrainedPotential(ofstream *out, atom **&PermutationMap, int startstep, int endstep, bool IsAngstroem)
1205{
1206 double Potential, OldPotential, OlderPotential;
1207 PermutationMap = (atom **) Malloc(AtomCount*sizeof(atom *), "molecule::MinimiseConstrainedPotential: **PermutationMap");
1208 DistanceMap **DistanceList = (DistanceMap **) Malloc(AtomCount*sizeof(DistanceMap *), "molecule::MinimiseConstrainedPotential: **DistanceList");
1209 DistanceMap::iterator *DistanceIterators = (DistanceMap::iterator *) Malloc(AtomCount*sizeof(DistanceMap::iterator), "molecule::MinimiseConstrainedPotential: *DistanceIterators");
1210 int *DoubleList = (int *) Malloc(AtomCount*sizeof(int), "molecule::MinimiseConstrainedPotential: *DoubleList");
1211 DistanceMap::iterator *StepList = (DistanceMap::iterator *) Malloc(AtomCount*sizeof(DistanceMap::iterator), "molecule::MinimiseConstrainedPotential: *StepList");
1212 double constants[3];
1213 int round;
1214 atom *Walker = NULL, *Runner = NULL, *Sprinter = NULL;
1215 DistanceMap::iterator Rider, Strider;
1216
1217 /// Minimise the potential
1218 // set Lagrange multiplier constants
1219 constants[0] = 10.;
1220 constants[1] = 1.;
1221 constants[2] = 1e+7; // just a huge penalty
1222 // generate the distance list
1223 *out << Verbose(1) << "Creating the distance list ... " << endl;
1224 for (int i=AtomCount; i--;) {
1225 DoubleList[i] = 0; // stores for how many atoms in startstep this atom is a target in endstep
1226 DistanceList[i] = new DistanceMap; // is the distance sorted target list per atom
1227 DistanceList[i]->clear();
1228 }
1229 *out << Verbose(1) << "Filling the distance list ... " << endl;
1230 Walker = start;
1231 while (Walker->next != end) {
1232 Walker = Walker->next;
1233 Runner = start;
1234 while(Runner->next != end) {
1235 Runner = Runner->next;
1236 DistanceList[Walker->nr]->insert( DistancePair(Trajectories[Walker].R.at(startstep).Distance(&Trajectories[Runner].R.at(endstep)), Runner) );
1237 }
1238 }
1239 // create the initial PermutationMap (source -> target)
1240 Walker = start;
1241 while (Walker->next != end) {
1242 Walker = Walker->next;
1243 StepList[Walker->nr] = DistanceList[Walker->nr]->begin(); // stores the step to the next iterator that could be a possible next target
1244 PermutationMap[Walker->nr] = DistanceList[Walker->nr]->begin()->second; // always pick target with the smallest distance
1245 DoubleList[DistanceList[Walker->nr]->begin()->second->nr]++; // increase this target's source count (>1? not injective)
1246 DistanceIterators[Walker->nr] = DistanceList[Walker->nr]->begin(); // and remember which one we picked
1247 *out << *Walker << " starts with distance " << DistanceList[Walker->nr]->begin()->first << "." << endl;
1248 }
1249 *out << Verbose(1) << "done." << endl;
1250 // make the PermutationMap injective by checking whether we have a non-zero constants[2] term in it
1251 *out << Verbose(1) << "Making the PermutationMap injective ... " << endl;
1252 Walker = start;
1253 DistanceMap::iterator NewBase;
1254 OldPotential = fabs(ConstrainedPotential(out, PermutationMap, startstep, endstep, constants, IsAngstroem));
1255 while ((OldPotential) > constants[2]) {
1256 PrintPermutationMap(out, PermutationMap, AtomCount);
1257 Walker = Walker->next;
1258 if (Walker == end) // round-robin at the end
1259 Walker = start->next;
1260 if (DoubleList[DistanceIterators[Walker->nr]->second->nr] <= 1) // no need to make those injective that aren't
1261 continue;
1262 // now, try finding a new one
1263 NewBase = DistanceIterators[Walker->nr]; // store old base
1264 do {
1265 NewBase++; // take next further distance in distance to targets list that's a target of no one
1266 } while ((DoubleList[NewBase->second->nr] != 0) && (NewBase != DistanceList[Walker->nr]->end()));
1267 if (NewBase != DistanceList[Walker->nr]->end()) {
1268 PermutationMap[Walker->nr] = NewBase->second;
1269 Potential = fabs(ConstrainedPotential(out, PermutationMap, startstep, endstep, constants, IsAngstroem));
1270 if (Potential > OldPotential) { // undo
1271 PermutationMap[Walker->nr] = DistanceIterators[Walker->nr]->second;
1272 } else { // do
1273 DoubleList[DistanceIterators[Walker->nr]->second->nr]--; // decrease the old entry in the doubles list
1274 DoubleList[NewBase->second->nr]++; // increase the old entry in the doubles list
1275 DistanceIterators[Walker->nr] = NewBase;
1276 OldPotential = Potential;
1277 *out << Verbose(3) << "Found a new permutation, new potential is " << OldPotential << "." << endl;
1278 }
1279 }
1280 }
1281 for (int i=AtomCount; i--;) // now each single entry in the DoubleList should be <=1
1282 if (DoubleList[i] > 1) {
1283 cerr << "Failed to create an injective PermutationMap!" << endl;
1284 exit(1);
1285 }
1286 *out << Verbose(1) << "done." << endl;
1287 Free((void **)&DoubleList, "molecule::MinimiseConstrainedPotential: *DoubleList");
1288 // argument minimise the constrained potential in this injective PermutationMap
1289 *out << Verbose(1) << "Argument minimising the PermutationMap, at current potential " << OldPotential << " ... " << endl;
1290 OldPotential = 1e+10;
1291 round = 0;
1292 do {
1293 *out << "Starting round " << ++round << " ... " << endl;
1294 OlderPotential = OldPotential;
1295 do {
1296 Walker = start;
1297 while (Walker->next != end) { // pick one
1298 Walker = Walker->next;
1299 PrintPermutationMap(out, PermutationMap, AtomCount);
1300 Sprinter = DistanceIterators[Walker->nr]->second; // store initial partner
1301 Strider = DistanceIterators[Walker->nr]; //remember old iterator
1302 DistanceIterators[Walker->nr] = StepList[Walker->nr];
1303 if (DistanceIterators[Walker->nr] == DistanceList[Walker->nr]->end()) {// stop, before we run through the list and still on
1304 DistanceIterators[Walker->nr] == DistanceList[Walker->nr]->begin();
1305 break;
1306 }
1307 //*out << Verbose(2) << "Current Walker: " << *Walker << " with old/next candidate " << *Sprinter << "/" << *DistanceIterators[Walker->nr]->second << "." << endl;
1308 // find source of the new target
1309 Runner = start->next;
1310 while(Runner != end) { // find the source whose toes we might be stepping on (Walker's new target should be in use by another already)
1311 if (PermutationMap[Runner->nr] == DistanceIterators[Walker->nr]->second) {
1312 //*out << Verbose(2) << "Found the corresponding owner " << *Runner << " to " << *PermutationMap[Runner->nr] << "." << endl;
1313 break;
1314 }
1315 Runner = Runner->next;
1316 }
1317 if (Runner != end) { // we found the other source
1318 // then look in its distance list for Sprinter
1319 Rider = DistanceList[Runner->nr]->begin();
1320 for (; Rider != DistanceList[Runner->nr]->end(); Rider++)
1321 if (Rider->second == Sprinter)
1322 break;
1323 if (Rider != DistanceList[Runner->nr]->end()) { // if we have found one
1324 //*out << Verbose(2) << "Current Other: " << *Runner << " with old/next candidate " << *PermutationMap[Runner->nr] << "/" << *Rider->second << "." << endl;
1325 // exchange both
1326 PermutationMap[Walker->nr] = DistanceIterators[Walker->nr]->second; // put next farther distance into PermutationMap
1327 PermutationMap[Runner->nr] = Sprinter; // and hand the old target to its respective owner
1328 PrintPermutationMap(out, PermutationMap, AtomCount);
1329 // calculate the new potential
1330 //*out << Verbose(2) << "Checking new potential ..." << endl;
1331 Potential = ConstrainedPotential(out, PermutationMap, startstep, endstep, constants, IsAngstroem);
1332 if (Potential > OldPotential) { // we made everything worse! Undo ...
1333 //*out << Verbose(3) << "Nay, made the potential worse: " << Potential << " vs. " << OldPotential << "!" << endl;
1334 //*out << Verbose(3) << "Setting " << *Runner << "'s source to " << *DistanceIterators[Runner->nr]->second << "." << endl;
1335 // Undo for Runner (note, we haven't moved the iteration yet, we may use this)
1336 PermutationMap[Runner->nr] = DistanceIterators[Runner->nr]->second;
1337 // Undo for Walker
1338 DistanceIterators[Walker->nr] = Strider; // take next farther distance target
1339 //*out << Verbose(3) << "Setting " << *Walker << "'s source to " << *DistanceIterators[Walker->nr]->second << "." << endl;
1340 PermutationMap[Walker->nr] = DistanceIterators[Walker->nr]->second;
1341 } else {
1342 DistanceIterators[Runner->nr] = Rider; // if successful also move the pointer in the iterator list
1343 *out << Verbose(3) << "Found a better permutation, new potential is " << Potential << " vs." << OldPotential << "." << endl;
1344 OldPotential = Potential;
1345 }
1346 if (Potential > constants[2]) {
1347 cerr << "ERROR: The two-step permutation procedure did not maintain injectivity!" << endl;
1348 exit(255);
1349 }
1350 //*out << endl;
1351 } else {
1352 cerr << "ERROR: " << *Runner << " was not the owner of " << *Sprinter << "!" << endl;
1353 exit(255);
1354 }
1355 } else {
1356 PermutationMap[Walker->nr] = DistanceIterators[Walker->nr]->second; // new target has no source!
1357 }
1358 StepList[Walker->nr]++; // take next farther distance target
1359 }
1360 } while (Walker->next != end);
1361 } while ((OlderPotential - OldPotential) > 1e-3);
1362 *out << Verbose(1) << "done." << endl;
1363
1364
1365 /// free memory and return with evaluated potential
1366 for (int i=AtomCount; i--;)
1367 DistanceList[i]->clear();
1368 Free((void **)&DistanceList, "molecule::MinimiseConstrainedPotential: **DistanceList");
1369 Free((void **)&DistanceIterators, "molecule::MinimiseConstrainedPotential: *DistanceIterators");
1370 return ConstrainedPotential(out, PermutationMap, startstep, endstep, constants, IsAngstroem);
1371};
1372
1373/** Evaluates the (distance-related part) of the constrained potential for the constrained forces.
1374 * \param *out output stream for debugging
1375 * \param startstep current MD step giving initial position between which and \a endstep we perform the constrained MD (as further steps are always concatenated)
1376 * \param endstep step giving final position in constrained MD
1377 * \param **PermutationMap mapping between the atom label of the initial and the final configuration
1378 * \param *Force ForceMatrix containing force vectors from the external energy functional minimisation.
1379 * \todo the constant for the constrained potential distance part is hard-coded independently of the hard-coded value in MinimiseConstrainedPotential()
1380 */
1381void molecule::EvaluateConstrainedForces(ofstream *out, int startstep, int endstep, atom **PermutationMap, ForceMatrix *Force)
1382{
1383 double constant = 10.;
1384 atom *Walker = NULL, *Sprinter = NULL;
1385
1386 /// evaluate forces (only the distance to target dependent part) with the final PermutationMap
1387 *out << Verbose(1) << "Calculating forces and adding onto ForceMatrix ... " << endl;
1388 Walker = start;
1389 while (Walker->next != NULL) {
1390 Walker = Walker->next;
1391 Sprinter = PermutationMap[Walker->nr];
1392 // set forces
1393 for (int i=NDIM;i++;)
1394 Force->Matrix[0][Walker->nr][5+i] += 2.*constant*sqrt(Trajectories[Walker].R.at(startstep).Distance(&Trajectories[Sprinter].R.at(endstep)));
1395 }
1396 *out << Verbose(1) << "done." << endl;
1397};
1398
1399/** Performs a linear interpolation between two desired atomic configurations with a given number of steps.
1400 * Note, step number is config::MaxOuterStep
1401 * \param *out output stream for debugging
1402 * \param startstep stating initial configuration in molecule::Trajectories
1403 * \param endstep stating final configuration in molecule::Trajectories
1404 * \param &config configuration structure
1405 * \return true - success in writing step files, false - error writing files or only one step in molecule::Trajectories
1406 */
1407bool molecule::LinearInterpolationBetweenConfiguration(ofstream *out, int startstep, int endstep, const char *prefix, config &configuration)
1408{
1409 bool status = true;
1410 int MaxSteps = configuration.MaxOuterStep;
1411 MoleculeListClass *MoleculePerStep = new MoleculeListClass(MaxSteps+1, AtomCount);
1412 // Get the Permutation Map by MinimiseConstrainedPotential
1413 atom **PermutationMap = NULL;
1414 atom *Walker = NULL, *Sprinter = NULL;
1415 MinimiseConstrainedPotential(out, PermutationMap, startstep, endstep, configuration.GetIsAngstroem());
1416
1417 // check whether we have sufficient space in Trajectories for each atom
1418 Walker = start;
1419 while (Walker->next != end) {
1420 Walker = Walker->next;
1421 if (Trajectories[Walker].R.size() <= (unsigned int)(MaxSteps)) {
1422 //cout << "Increasing size for trajectory array of " << keyword << " to " << (MaxSteps+1) << "." << endl;
1423 Trajectories[Walker].R.resize(MaxSteps+1);
1424 Trajectories[Walker].U.resize(MaxSteps+1);
1425 Trajectories[Walker].F.resize(MaxSteps+1);
1426 }
1427 }
1428 // push endstep to last one
1429 Walker = start;
1430 while (Walker->next != end) { // remove the endstep (is now the very last one)
1431 Walker = Walker->next;
1432 for (int n=NDIM;n--;) {
1433 Trajectories[Walker].R.at(MaxSteps).x[n] = Trajectories[Walker].R.at(endstep).x[n];
1434 Trajectories[Walker].U.at(MaxSteps).x[n] = Trajectories[Walker].U.at(endstep).x[n];
1435 Trajectories[Walker].F.at(MaxSteps).x[n] = Trajectories[Walker].F.at(endstep).x[n];
1436 }
1437 }
1438 endstep = MaxSteps;
1439
1440 // go through all steps and add the molecular configuration to the list and to the Trajectories of \a this molecule
1441 *out << Verbose(1) << "Filling intermediate " << MaxSteps << " steps with MDSteps of " << MDSteps << "." << endl;
1442 for (int step = 0; step <= MaxSteps; step++) {
1443 MoleculePerStep->ListOfMolecules[step] = new molecule(elemente);
1444 Walker = start;
1445 while (Walker->next != end) {
1446 Walker = Walker->next;
1447 // add to molecule list
1448 Sprinter = MoleculePerStep->ListOfMolecules[step]->AddCopyAtom(Walker);
1449 for (int n=NDIM;n--;) {
1450 Sprinter->x.x[n] = Trajectories[Walker].R.at(startstep).x[n] + (Trajectories[PermutationMap[Walker->nr]].R.at(endstep).x[n] - Trajectories[Walker].R.at(startstep).x[n])*((double)step/(double)MaxSteps);
1451 // add to Trajectories
1452 //*out << Verbose(3) << step << ">=" << MDSteps-1 << endl;
1453 if (step < MaxSteps) {
1454 Trajectories[Walker].R.at(step).x[n] = Trajectories[Walker].R.at(startstep).x[n] + (Trajectories[PermutationMap[Walker->nr]].R.at(endstep).x[n] - Trajectories[Walker].R.at(startstep).x[n])*((double)step/(double)MaxSteps);
1455 Trajectories[Walker].U.at(step).x[n] = 0.;
1456 Trajectories[Walker].F.at(step).x[n] = 0.;
1457 }
1458 }
1459 }
1460 }
1461 MDSteps = MaxSteps+1; // otherwise new Trajectories' points aren't stored on save&exit
1462
1463 // store the list to single step files
1464 int *SortIndex = (int *) Malloc(AtomCount*sizeof(int), "molecule::LinearInterpolationBetweenConfiguration: *SortIndex");
1465 for (int i=AtomCount; i--; )
1466 SortIndex[i] = i;
1467 status = MoleculePerStep->OutputConfigForListOfFragments(out, "ConstrainedStep", &configuration, SortIndex, false, false);
1468
1469 // free and return
1470 Free((void **)&PermutationMap, "molecule::MinimiseConstrainedPotential: *PermutationMap");
1471 delete(MoleculePerStep);
1472 return status;
1473};
1474
1475/** Parses nuclear forces from file and performs Verlet integration.
1476 * Note that we assume the parsed forces to be in atomic units (hence, if coordinates are in angstroem, we
1477 * have to transform them).
1478 * This adds a new MD step to the config file.
1479 * \param *out output stream for debugging
1480 * \param *file filename
1481 * \param config structure with config::Deltat, config::IsAngstroem, config::DoConstrained
1482 * \param delta_t time step width in atomic units
1483 * \param IsAngstroem whether coordinates are in angstroem (true) or bohrradius (false)
1484 * \param DoConstrained whether we perform a constrained (>0, target step in molecule::trajectories) or unconstrained (0) molecular dynamics, \sa molecule::MinimiseConstrainedPotential()
1485 * \return true - file found and parsed, false - file not found or imparsable
1486 * \todo This is not yet checked if it is correctly working with DoConstrained set to true.
1487 */
1488bool molecule::VerletForceIntegration(ofstream *out, char *file, config &configuration)
1489{
1490 atom *walker = NULL;
1491 int AtomNo;
1492 ifstream input(file);
1493 string token;
1494 stringstream item;
1495 double a, IonMass, Vector[NDIM], ConstrainedPotentialEnergy, ActualTemp;
1496 ForceMatrix Force;
1497
1498 CountElements(); // make sure ElementsInMolecule is up to date
1499
1500 // check file
1501 if (input == NULL) {
1502 return false;
1503 } else {
1504 // parse file into ForceMatrix
1505 if (!Force.ParseMatrix(file, 0,0,0)) {
1506 cerr << "Could not parse Force Matrix file " << file << "." << endl;
1507 return false;
1508 }
1509 if (Force.RowCounter[0] != AtomCount) {
1510 cerr << "Mismatch between number of atoms in file " << Force.RowCounter[0] << " and in molecule " << AtomCount << "." << endl;
1511 return false;
1512 }
1513 // correct Forces
1514 for(int d=0;d<NDIM;d++)
1515 Vector[d] = 0.;
1516 for(int i=0;i<AtomCount;i++)
1517 for(int d=0;d<NDIM;d++) {
1518 Vector[d] += Force.Matrix[0][i][d+5];
1519 }
1520 for(int i=0;i<AtomCount;i++)
1521 for(int d=0;d<NDIM;d++) {
1522 Force.Matrix[0][i][d+5] -= Vector[d]/(double)AtomCount;
1523 }
1524 // solve a constrained potential if we are meant to
1525 if (configuration.DoConstrainedMD) {
1526 // calculate forces and potential
1527 atom **PermutationMap = NULL;
1528 ConstrainedPotentialEnergy = MinimiseConstrainedPotential(out, PermutationMap,configuration.DoConstrainedMD, 0, configuration.GetIsAngstroem());
1529 EvaluateConstrainedForces(out, configuration.DoConstrainedMD, 0, PermutationMap, &Force);
1530 Free((void **)&PermutationMap, "molecule::MinimiseConstrainedPotential: *PermutationMap");
1531 }
1532
1533 // and perform Verlet integration for each atom with position, velocity and force vector
1534 walker = start;
1535 while (walker->next != end) { // go through every atom of this element
1536 walker = walker->next;
1537 //a = configuration.Deltat*0.5/walker->type->mass; // (F+F_old)/2m = a and thus: v = (F+F_old)/2m * t = (F + F_old) * a
1538 // check size of vectors
1539 if (Trajectories[walker].R.size() <= (unsigned int)(MDSteps)) {
1540 //out << "Increasing size for trajectory array of " << *walker << " to " << (size+10) << "." << endl;
1541 Trajectories[walker].R.resize(MDSteps+10);
1542 Trajectories[walker].U.resize(MDSteps+10);
1543 Trajectories[walker].F.resize(MDSteps+10);
1544 }
1545
1546 // Update R (and F)
1547 for (int d=0; d<NDIM; d++) {
1548 Trajectories[walker].F.at(MDSteps).x[d] = -Force.Matrix[0][walker->nr][d+5]*(configuration.GetIsAngstroem() ? AtomicLengthToAngstroem : 1.);
1549 Trajectories[walker].R.at(MDSteps).x[d] = Trajectories[walker].R.at(MDSteps-1).x[d];
1550 Trajectories[walker].R.at(MDSteps).x[d] += configuration.Deltat*(Trajectories[walker].U.at(MDSteps-1).x[d]); // s(t) = s(0) + v * deltat + 1/2 a * deltat^2
1551 Trajectories[walker].R.at(MDSteps).x[d] += 0.5*configuration.Deltat*configuration.Deltat*(Trajectories[walker].F.at(MDSteps).x[d]/walker->type->mass); // F = m * a and s = 0.5 * F/m * t^2 = F * a * t
1552 }
1553 // Update U
1554 for (int d=0; d<NDIM; d++) {
1555 Trajectories[walker].U.at(MDSteps).x[d] = Trajectories[walker].U.at(MDSteps-1).x[d];
1556 Trajectories[walker].U.at(MDSteps).x[d] += configuration.Deltat * (Trajectories[walker].F.at(MDSteps).x[d]+Trajectories[walker].F.at(MDSteps-1).x[d]/walker->type->mass); // v = F/m * t
1557 }
1558// out << "Integrated position&velocity of step " << (MDSteps) << ": (";
1559// for (int d=0;d<NDIM;d++)
1560// out << Trajectories[walker].R.at(MDSteps).x[d] << " "; // next step
1561// out << ")\t(";
1562// for (int d=0;d<NDIM;d++)
1563// cout << Trajectories[walker].U.at(MDSteps).x[d] << " "; // next step
1564// out << ")" << endl;
1565 // next atom
1566 }
1567 }
1568 // correct velocities (rather momenta) so that center of mass remains motionless
1569 for(int d=0;d<NDIM;d++)
1570 Vector[d] = 0.;
1571 IonMass = 0.;
1572 walker = start;
1573 while (walker->next != end) { // go through every atom
1574 walker = walker->next;
1575 IonMass += walker->type->mass; // sum up total mass
1576 for(int d=0;d<NDIM;d++) {
1577 Vector[d] += Trajectories[walker].U.at(MDSteps).x[d]*walker->type->mass;
1578 }
1579 }
1580 for(int d=0;d<NDIM;d++)
1581 Vector[d] /= IonMass;
1582 ActualTemp = 0.;
1583 walker = start;
1584 while (walker->next != end) { // go through every atom of this element
1585 walker = walker->next;
1586 for(int d=0;d<NDIM;d++) {
1587 Trajectories[walker].U.at(MDSteps).x[d] -= Vector[d];
1588 ActualTemp += 0.5 * walker->type->mass * Trajectories[walker].U.at(MDSteps).x[d] * Trajectories[walker].U.at(MDSteps).x[d];
1589 }
1590 }
1591 Thermostats(configuration, ActualTemp, Berendsen);
1592 MDSteps++;
1593
1594
1595 // exit
1596 return true;
1597};
1598
1599
1600/** Implementation of various thermostats.
1601 * All these thermostats apply an additional force which has the following forms:
1602 * -# Woodcock
1603 * \f$p_i \rightarrow \sqrt{\frac{T_0}{T}} \cdot p_i\f$
1604 * -# Gaussian
1605 * \f$ \frac{ \sum_i \frac{p_i}{m_i} \frac{\partial V}{\partial q_i}} {\sum_i \frac{p^2_i}{m_i}} \cdot p_i\f$
1606 * -# Langevin
1607 * \f$p_{i,n} \rightarrow \sqrt{1-\alpha^2} p_{i,0} + \alpha p_r\f$
1608 * -# Berendsen
1609 * \f$p_i \rightarrow \left [ 1+ \frac{\delta t}{\tau_T} \left ( \frac{T_0}{T} \right ) \right ]^{\frac{1}{2}} \cdot p_i\f$
1610 * -# Nose-Hoover
1611 * \f$\zeta p_i \f$ with \f$\frac{\partial \zeta}{\partial t} = \frac{1}{M_s} \left ( \sum^N_{i=1} \frac{p_i^2}{m_i} - g k_B T \right )\f$
1612 * These Thermostats either simply rescale the velocities, thus this function should be called after ion velocities have been updated, and/or
1613 * have a constraint force acting additionally on the ions. In the latter case, the ion speeds have to be modified
1614 * belatedly and the constraint force set.
1615 * \param *P Problem at hand
1616 * \param i which of the thermostats to take: 0 - none, 1 - Woodcock, 2 - Gaussian, 3 - Langevin, 4 - Berendsen, 5 - Nose-Hoover
1617 * \sa InitThermostat()
1618 */
1619void molecule::Thermostats(config &configuration, double ActualTemp, int Thermostat)
1620{
1621 double ekin = 0.;
1622 double E = 0., G = 0.;
1623 double delta_alpha = 0.;
1624 double ScaleTempFactor;
1625 double sigma;
1626 double IonMass;
1627 int d;
1628 gsl_rng * r;
1629 const gsl_rng_type * T;
1630 double *U = NULL, *F = NULL, FConstraint[NDIM];
1631 atom *walker = NULL;
1632
1633 // calculate scale configuration
1634 ScaleTempFactor = configuration.TargetTemp/ActualTemp;
1635
1636 // differentating between the various thermostats
1637 switch(Thermostat) {
1638 case None:
1639 cout << Verbose(2) << "Applying no thermostat..." << endl;
1640 break;
1641 case Woodcock:
1642 if ((configuration.ScaleTempStep > 0) && ((MDSteps-1) % configuration.ScaleTempStep == 0)) {
1643 cout << Verbose(2) << "Applying Woodcock thermostat..." << endl;
1644 walker = start;
1645 while (walker->next != end) { // go through every atom of this element
1646 walker = walker->next;
1647 IonMass = walker->type->mass;
1648 U = Trajectories[walker].U.at(MDSteps).x;
1649 if (walker->FixedIon == 0) // even FixedIon moves, only not by other's forces
1650 for (d=0; d<NDIM; d++) {
1651 U[d] *= sqrt(ScaleTempFactor);
1652 ekin += 0.5*IonMass * U[d]*U[d];
1653 }
1654 }
1655 }
1656 break;
1657 case Gaussian:
1658 cout << Verbose(2) << "Applying Gaussian thermostat..." << endl;
1659 walker = start;
1660 while (walker->next != end) { // go through every atom of this element
1661 walker = walker->next;
1662 IonMass = walker->type->mass;
1663 U = Trajectories[walker].U.at(MDSteps).x;
1664 F = Trajectories[walker].F.at(MDSteps).x;
1665 if (walker->FixedIon == 0) // even FixedIon moves, only not by other's forces
1666 for (d=0; d<NDIM; d++) {
1667 G += U[d] * F[d];
1668 E += U[d]*U[d]*IonMass;
1669 }
1670 }
1671 cout << Verbose(1) << "Gaussian Least Constraint constant is " << G/E << "." << endl;
1672 walker = start;
1673 while (walker->next != end) { // go through every atom of this element
1674 walker = walker->next;
1675 IonMass = walker->type->mass;
1676 U = Trajectories[walker].U.at(MDSteps).x;
1677 F = Trajectories[walker].F.at(MDSteps).x;
1678 if (walker->FixedIon == 0) // even FixedIon moves, only not by other's forces
1679 for (d=0; d<NDIM; d++) {
1680 FConstraint[d] = (G/E) * (U[d]*IonMass);
1681 U[d] += configuration.Deltat/IonMass * (FConstraint[d]);
1682 ekin += IonMass * U[d]*U[d];
1683 }
1684 }
1685 break;
1686 case Langevin:
1687 cout << Verbose(2) << "Applying Langevin thermostat..." << endl;
1688 // init random number generator
1689 gsl_rng_env_setup();
1690 T = gsl_rng_default;
1691 r = gsl_rng_alloc (T);
1692 // Go through each ion
1693 walker = start;
1694 while (walker->next != end) { // go through every atom of this element
1695 walker = walker->next;
1696 IonMass = walker->type->mass;
1697 sigma = sqrt(configuration.TargetTemp/IonMass); // sigma = (k_b T)/m (Hartree/atomicmass = atomiclength/atomictime)
1698 U = Trajectories[walker].U.at(MDSteps).x;
1699 F = Trajectories[walker].F.at(MDSteps).x;
1700 if (walker->FixedIon == 0) { // even FixedIon moves, only not by other's forces
1701 // throw a dice to determine whether it gets hit by a heat bath particle
1702 if (((((rand()/(double)RAND_MAX))*configuration.TempFrequency) < 1.)) {
1703 cout << Verbose(3) << "Particle " << *walker << " was hit (sigma " << sigma << "): " << sqrt(U[0]*U[0]+U[1]*U[1]+U[2]*U[2]) << " -> ";
1704 // pick three random numbers from a Boltzmann distribution around the desired temperature T for each momenta axis
1705 for (d=0; d<NDIM; d++) {
1706 U[d] = gsl_ran_gaussian (r, sigma);
1707 }
1708 cout << sqrt(U[0]*U[0]+U[1]*U[1]+U[2]*U[2]) << endl;
1709 }
1710 for (d=0; d<NDIM; d++)
1711 ekin += 0.5*IonMass * U[d]*U[d];
1712 }
1713 }
1714 break;
1715 case Berendsen:
1716 cout << Verbose(2) << "Applying Berendsen-VanGunsteren thermostat..." << endl;
1717 walker = start;
1718 while (walker->next != end) { // go through every atom of this element
1719 walker = walker->next;
1720 IonMass = walker->type->mass;
1721 U = Trajectories[walker].U.at(MDSteps).x;
1722 F = Trajectories[walker].F.at(MDSteps).x;
1723 if (walker->FixedIon == 0) { // even FixedIon moves, only not by other's forces
1724 for (d=0; d<NDIM; d++) {
1725 U[d] *= sqrt(1+(configuration.Deltat/configuration.TempFrequency)*(ScaleTempFactor-1));
1726 ekin += 0.5*IonMass * U[d]*U[d];
1727 }
1728 }
1729 }
1730 break;
1731 case NoseHoover:
1732 cout << Verbose(2) << "Applying Nose-Hoover thermostat..." << endl;
1733 // dynamically evolve alpha (the additional degree of freedom)
1734 delta_alpha = 0.;
1735 walker = start;
1736 while (walker->next != end) { // go through every atom of this element
1737 walker = walker->next;
1738 IonMass = walker->type->mass;
1739 U = Trajectories[walker].U.at(MDSteps).x;
1740 if (walker->FixedIon == 0) { // even FixedIon moves, only not by other's forces
1741 for (d=0; d<NDIM; d++) {
1742 delta_alpha += U[d]*U[d]*IonMass;
1743 }
1744 }
1745 }
1746 delta_alpha = (delta_alpha - (3.*AtomCount+1.) * configuration.TargetTemp)/(configuration.HooverMass*Units2Electronmass);
1747 configuration.alpha += delta_alpha*configuration.Deltat;
1748 cout << Verbose(3) << "alpha = " << delta_alpha << " * " << configuration.Deltat << " = " << configuration.alpha << "." << endl;
1749 // apply updated alpha as additional force
1750 walker = start;
1751 while (walker->next != end) { // go through every atom of this element
1752 walker = walker->next;
1753 IonMass = walker->type->mass;
1754 U = Trajectories[walker].U.at(MDSteps).x;
1755 if (walker->FixedIon == 0) { // even FixedIon moves, only not by other's forces
1756 for (d=0; d<NDIM; d++) {
1757 FConstraint[d] = - configuration.alpha * (U[d] * IonMass);
1758 U[d] += configuration.Deltat/IonMass * (FConstraint[d]);
1759 ekin += (0.5*IonMass) * U[d]*U[d];
1760 }
1761 }
1762 }
1763 break;
1764 }
1765 cout << Verbose(1) << "Kinetic energy is " << ekin << "." << endl;
1766};
1767
1768/** Align all atoms in such a manner that given vector \a *n is along z axis.
1769 * \param n[] alignment vector.
1770 */
1771void molecule::Align(Vector *n)
1772{
1773 atom *ptr = start;
1774 double alpha, tmp;
1775 Vector z_axis;
1776 z_axis.x[0] = 0.;
1777 z_axis.x[1] = 0.;
1778 z_axis.x[2] = 1.;
1779
1780 // rotate on z-x plane
1781 cout << Verbose(0) << "Begin of Aligning all atoms." << endl;
1782 alpha = atan(-n->x[0]/n->x[2]);
1783 cout << Verbose(1) << "Z-X-angle: " << alpha << " ... ";
1784 while (ptr->next != end) {
1785 ptr = ptr->next;
1786 tmp = ptr->x.x[0];
1787 ptr->x.x[0] = cos(alpha) * tmp + sin(alpha) * ptr->x.x[2];
1788 ptr->x.x[2] = -sin(alpha) * tmp + cos(alpha) * ptr->x.x[2];
1789 for (int j=0;j<MDSteps;j++) {
1790 tmp = Trajectories[ptr].R.at(j).x[0];
1791 Trajectories[ptr].R.at(j).x[0] = cos(alpha) * tmp + sin(alpha) * Trajectories[ptr].R.at(j).x[2];
1792 Trajectories[ptr].R.at(j).x[2] = -sin(alpha) * tmp + cos(alpha) * Trajectories[ptr].R.at(j).x[2];
1793 }
1794 }
1795 // rotate n vector
1796 tmp = n->x[0];
1797 n->x[0] = cos(alpha) * tmp + sin(alpha) * n->x[2];
1798 n->x[2] = -sin(alpha) * tmp + cos(alpha) * n->x[2];
1799 cout << Verbose(1) << "alignment vector after first rotation: ";
1800 n->Output((ofstream *)&cout);
1801 cout << endl;
1802
1803 // rotate on z-y plane
1804 ptr = start;
1805 alpha = atan(-n->x[1]/n->x[2]);
1806 cout << Verbose(1) << "Z-Y-angle: " << alpha << " ... ";
1807 while (ptr->next != end) {
1808 ptr = ptr->next;
1809 tmp = ptr->x.x[1];
1810 ptr->x.x[1] = cos(alpha) * tmp + sin(alpha) * ptr->x.x[2];
1811 ptr->x.x[2] = -sin(alpha) * tmp + cos(alpha) * ptr->x.x[2];
1812 for (int j=0;j<MDSteps;j++) {
1813 tmp = Trajectories[ptr].R.at(j).x[1];
1814 Trajectories[ptr].R.at(j).x[1] = cos(alpha) * tmp + sin(alpha) * Trajectories[ptr].R.at(j).x[2];
1815 Trajectories[ptr].R.at(j).x[2] = -sin(alpha) * tmp + cos(alpha) * Trajectories[ptr].R.at(j).x[2];
1816 }
1817 }
1818 // rotate n vector (for consistency check)
1819 tmp = n->x[1];
1820 n->x[1] = cos(alpha) * tmp + sin(alpha) * n->x[2];
1821 n->x[2] = -sin(alpha) * tmp + cos(alpha) * n->x[2];
1822
1823 cout << Verbose(1) << "alignment vector after second rotation: ";
1824 n->Output((ofstream *)&cout);
1825 cout << Verbose(1) << endl;
1826 cout << Verbose(0) << "End of Aligning all atoms." << endl;
1827};
1828
1829/** Removes atom from molecule list.
1830 * \param *pointer atom to be removed
1831 * \return true - succeeded, false - atom not found in list
1832 */
1833bool molecule::RemoveAtom(atom *pointer)
1834{
1835 if (ElementsInMolecule[pointer->type->Z] != 0) // this would indicate an error
1836 ElementsInMolecule[pointer->type->Z]--; // decrease number of atom of this element
1837 else
1838 cerr << "ERROR: Atom " << pointer->Name << " is of element " << pointer->type->Z << " but the entry in the table of the molecule is 0!" << endl;
1839 if (ElementsInMolecule[pointer->type->Z] == 0) // was last atom of this element?
1840 ElementCount--;
1841 Trajectories.erase(pointer);
1842 return remove(pointer, start, end);
1843};
1844
1845/** Removes every atom from molecule list.
1846 * \return true - succeeded, false - atom not found in list
1847 */
1848bool molecule::CleanupMolecule()
1849{
1850 return (cleanup(start,end) && cleanup(first,last));
1851};
1852
1853/** Finds an atom specified by its continuous number.
1854 * \param Nr number of atom withim molecule
1855 * \return pointer to atom or NULL
1856 */
1857atom * molecule::FindAtom(int Nr) const{
1858 atom * walker = find(&Nr, start,end);
1859 if (walker != NULL) {
1860 //cout << Verbose(0) << "Found Atom Nr. " << walker->nr << endl;
1861 return walker;
1862 } else {
1863 cout << Verbose(0) << "Atom not found in list." << endl;
1864 return NULL;
1865 }
1866};
1867
1868/** Asks for atom number, and checks whether in list.
1869 * \param *text question before entering
1870 */
1871atom * molecule::AskAtom(string text)
1872{
1873 int No;
1874 atom *ion = NULL;
1875 do {
1876 //cout << Verbose(0) << "============Atom list==========================" << endl;
1877 //mol->Output((ofstream *)&cout);
1878 //cout << Verbose(0) << "===============================================" << endl;
1879 cout << Verbose(0) << text;
1880 cin >> No;
1881 ion = this->FindAtom(No);
1882 } while (ion == NULL);
1883 return ion;
1884};
1885
1886/** Checks if given coordinates are within cell volume.
1887 * \param *x array of coordinates
1888 * \return true - is within, false - out of cell
1889 */
1890bool molecule::CheckBounds(const Vector *x) const
1891{
1892 bool result = true;
1893 int j =-1;
1894 for (int i=0;i<NDIM;i++) {
1895 j += i+1;
1896 result = result && ((x->x[i] >= 0) && (x->x[i] < cell_size[j]));
1897 }
1898 //return result;
1899 return true; /// probably not gonna use the check no more
1900};
1901
1902/** Calculates sum over least square distance to line hidden in \a *x.
1903 * \param *x offset and direction vector
1904 * \param *params pointer to lsq_params structure
1905 * \return \f$ sum_i^N | y_i - (a + t_i b)|^2\f$
1906 */
1907double LeastSquareDistance (const gsl_vector * x, void * params)
1908{
1909 double res = 0, t;
1910 Vector a,b,c,d;
1911 struct lsq_params *par = (struct lsq_params *)params;
1912 atom *ptr = par->mol->start;
1913
1914 // initialize vectors
1915 a.x[0] = gsl_vector_get(x,0);
1916 a.x[1] = gsl_vector_get(x,1);
1917 a.x[2] = gsl_vector_get(x,2);
1918 b.x[0] = gsl_vector_get(x,3);
1919 b.x[1] = gsl_vector_get(x,4);
1920 b.x[2] = gsl_vector_get(x,5);
1921 // go through all atoms
1922 while (ptr != par->mol->end) {
1923 ptr = ptr->next;
1924 if (ptr->type == ((struct lsq_params *)params)->type) { // for specific type
1925 c.CopyVector(&ptr->x); // copy vector to temporary one
1926 c.SubtractVector(&a); // subtract offset vector
1927 t = c.ScalarProduct(&b); // get direction parameter
1928 d.CopyVector(&b); // and create vector
1929 d.Scale(&t);
1930 c.SubtractVector(&d); // ... yielding distance vector
1931 res += d.ScalarProduct((const Vector *)&d); // add squared distance
1932 }
1933 }
1934 return res;
1935};
1936
1937/** By minimizing the least square distance gains alignment vector.
1938 * \bug this is not yet working properly it seems
1939 */
1940void molecule::GetAlignvector(struct lsq_params * par) const
1941{
1942 int np = 6;
1943
1944 const gsl_multimin_fminimizer_type *T =
1945 gsl_multimin_fminimizer_nmsimplex;
1946 gsl_multimin_fminimizer *s = NULL;
1947 gsl_vector *ss;
1948 gsl_multimin_function minex_func;
1949
1950 size_t iter = 0, i;
1951 int status;
1952 double size;
1953
1954 /* Initial vertex size vector */
1955 ss = gsl_vector_alloc (np);
1956
1957 /* Set all step sizes to 1 */
1958 gsl_vector_set_all (ss, 1.0);
1959
1960 /* Starting point */
1961 par->x = gsl_vector_alloc (np);
1962 par->mol = this;
1963
1964 gsl_vector_set (par->x, 0, 0.0); // offset
1965 gsl_vector_set (par->x, 1, 0.0);
1966 gsl_vector_set (par->x, 2, 0.0);
1967 gsl_vector_set (par->x, 3, 0.0); // direction
1968 gsl_vector_set (par->x, 4, 0.0);
1969 gsl_vector_set (par->x, 5, 1.0);
1970
1971 /* Initialize method and iterate */
1972 minex_func.f = &LeastSquareDistance;
1973 minex_func.n = np;
1974 minex_func.params = (void *)par;
1975
1976 s = gsl_multimin_fminimizer_alloc (T, np);
1977 gsl_multimin_fminimizer_set (s, &minex_func, par->x, ss);
1978
1979 do
1980 {
1981 iter++;
1982 status = gsl_multimin_fminimizer_iterate(s);
1983
1984 if (status)
1985 break;
1986
1987 size = gsl_multimin_fminimizer_size (s);
1988 status = gsl_multimin_test_size (size, 1e-2);
1989
1990 if (status == GSL_SUCCESS)
1991 {
1992 printf ("converged to minimum at\n");
1993 }
1994
1995 printf ("%5d ", (int)iter);
1996 for (i = 0; i < (size_t)np; i++)
1997 {
1998 printf ("%10.3e ", gsl_vector_get (s->x, i));
1999 }
2000 printf ("f() = %7.3f size = %.3f\n", s->fval, size);
2001 }
2002 while (status == GSL_CONTINUE && iter < 100);
2003
2004 for (i=0;i<(size_t)np;i++)
2005 gsl_vector_set(par->x, i, gsl_vector_get(s->x, i));
2006 //gsl_vector_free(par->x);
2007 gsl_vector_free(ss);
2008 gsl_multimin_fminimizer_free (s);
2009};
2010
2011/** Prints molecule to *out.
2012 * \param *out output stream
2013 */
2014bool molecule::Output(ofstream *out)
2015{
2016 element *runner;
2017 atom *walker = NULL;
2018 int ElementNo, AtomNo;
2019 CountElements();
2020
2021 if (out == NULL) {
2022 return false;
2023 } else {
2024 *out << "#Ion_TypeNr._Nr.R[0] R[1] R[2] MoveType (0 MoveIon, 1 FixedIon)" << endl;
2025 ElementNo = 0;
2026 runner = elemente->start;
2027 while (runner->next != elemente->end) { // go through every element
2028 runner = runner->next;
2029 if (ElementsInMolecule[runner->Z]) { // if this element got atoms
2030 ElementNo++;
2031 AtomNo = 0;
2032 walker = start;
2033 while (walker->next != end) { // go through every atom of this element
2034 walker = walker->next;
2035 if (walker->type == runner) { // if this atom fits to element
2036 AtomNo++;
2037 walker->Output(ElementNo, AtomNo, out); // removed due to trajectories
2038 }
2039 }
2040 }
2041 }
2042 return true;
2043 }
2044};
2045
2046/** Prints molecule with all atomic trajectory positions to *out.
2047 * \param *out output stream
2048 */
2049bool molecule::OutputTrajectories(ofstream *out)
2050{
2051 element *runner = NULL;
2052 atom *walker = NULL;
2053 int ElementNo, AtomNo;
2054 CountElements();
2055
2056 if (out == NULL) {
2057 return false;
2058 } else {
2059 for (int step = 0; step < MDSteps; step++) {
2060 if (step == 0) {
2061 *out << "#Ion_TypeNr._Nr.R[0] R[1] R[2] MoveType (0 MoveIon, 1 FixedIon)" << endl;
2062 } else {
2063 *out << "# ====== MD step " << step << " =========" << endl;
2064 }
2065 ElementNo = 0;
2066 runner = elemente->start;
2067 while (runner->next != elemente->end) { // go through every element
2068 runner = runner->next;
2069 if (ElementsInMolecule[runner->Z]) { // if this element got atoms
2070 ElementNo++;
2071 AtomNo = 0;
2072 walker = start;
2073 while (walker->next != end) { // go through every atom of this element
2074 walker = walker->next;
2075 if (walker->type == runner) { // if this atom fits to element
2076 AtomNo++;
2077 *out << "Ion_Type" << ElementNo << "_" << AtomNo << "\t" << fixed << setprecision(9) << showpoint;
2078 *out << Trajectories[walker].R.at(step).x[0] << "\t" << Trajectories[walker].R.at(step).x[1] << "\t" << Trajectories[walker].R.at(step).x[2];
2079 *out << "\t" << walker->FixedIon;
2080 if (Trajectories[walker].U.at(step).Norm() > MYEPSILON)
2081 *out << "\t" << scientific << setprecision(6) << Trajectories[walker].U.at(step).x[0] << "\t" << Trajectories[walker].U.at(step).x[1] << "\t" << Trajectories[walker].U.at(step).x[2] << "\t";
2082 if (Trajectories[walker].F.at(step).Norm() > MYEPSILON)
2083 *out << "\t" << scientific << setprecision(6) << Trajectories[walker].F.at(step).x[0] << "\t" << Trajectories[walker].F.at(step).x[1] << "\t" << Trajectories[walker].F.at(step).x[2] << "\t";
2084 *out << "\t# Number in molecule " << walker->nr << endl;
2085 }
2086 }
2087 }
2088 }
2089 }
2090 return true;
2091 }
2092};
2093
2094/** Outputs contents of molecule::ListOfBondsPerAtom.
2095 * \param *out output stream
2096 */
2097void molecule::OutputListOfBonds(ofstream *out) const
2098{
2099 *out << Verbose(2) << endl << "From Contents of ListOfBondsPerAtom, all non-hydrogen atoms:" << endl;
2100 atom *Walker = start;
2101 while (Walker->next != end) {
2102 Walker = Walker->next;
2103#ifdef ADDHYDROGEN
2104 if (Walker->type->Z != 1) { // regard only non-hydrogen
2105#endif
2106 *out << Verbose(2) << "Atom " << Walker->Name << " has Bonds: "<<endl;
2107 for(int j=0;j<NumberOfBondsPerAtom[Walker->nr];j++) {
2108 *out << Verbose(3) << *(ListOfBondsPerAtom)[Walker->nr][j] << endl;
2109 }
2110#ifdef ADDHYDROGEN
2111 }
2112#endif
2113 }
2114 *out << endl;
2115};
2116
2117/** Output of element before the actual coordination list.
2118 * \param *out stream pointer
2119 */
2120bool molecule::Checkout(ofstream *out) const
2121{
2122 return elemente->Checkout(out, ElementsInMolecule);
2123};
2124
2125/** Prints molecule with all its trajectories to *out as xyz file.
2126 * \param *out output stream
2127 */
2128bool molecule::OutputTrajectoriesXYZ(ofstream *out)
2129{
2130 atom *walker = NULL;
2131 int No = 0;
2132 time_t now;
2133
2134 now = time((time_t *)NULL); // Get the system time and put it into 'now' as 'calender time'
2135 walker = start;
2136 while (walker->next != end) { // go through every atom and count
2137 walker = walker->next;
2138 No++;
2139 }
2140 if (out != NULL) {
2141 for (int step=0;step<MDSteps;step++) {
2142 *out << No << "\n\tCreated by molecuilder, step " << step << ", on " << ctime(&now);
2143 walker = start;
2144 while (walker->next != end) { // go through every atom of this element
2145 walker = walker->next;
2146 *out << walker->type->symbol << "\t" << Trajectories[walker].R.at(step).x[0] << "\t" << Trajectories[walker].R.at(step).x[1] << "\t" << Trajectories[walker].R.at(step).x[2] << endl;
2147 }
2148 }
2149 return true;
2150 } else
2151 return false;
2152};
2153
2154/** Prints molecule to *out as xyz file.
2155* \param *out output stream
2156 */
2157bool molecule::OutputXYZ(ofstream *out) const
2158{
2159 atom *walker = NULL;
2160 int No = 0;
2161 time_t now;
2162
2163 now = time((time_t *)NULL); // Get the system time and put it into 'now' as 'calender time'
2164 walker = start;
2165 while (walker->next != end) { // go through every atom and count
2166 walker = walker->next;
2167 No++;
2168 }
2169 if (out != NULL) {
2170 *out << No << "\n\tCreated by molecuilder on " << ctime(&now);
2171 walker = start;
2172 while (walker->next != end) { // go through every atom of this element
2173 walker = walker->next;
2174 walker->OutputXYZLine(out);
2175 }
2176 return true;
2177 } else
2178 return false;
2179};
2180
2181/** Brings molecule::AtomCount and atom::*Name up-to-date.
2182 * \param *out output stream for debugging
2183 */
2184void molecule::CountAtoms(ofstream *out)
2185{
2186 int i = 0;
2187 atom *Walker = start;
2188 while (Walker->next != end) {
2189 Walker = Walker->next;
2190 i++;
2191 }
2192 if ((AtomCount == 0) || (i != AtomCount)) {
2193 *out << Verbose(3) << "Mismatch in AtomCount " << AtomCount << " and recounted number " << i << ", renaming all." << endl;
2194 AtomCount = i;
2195
2196 // count NonHydrogen atoms and give each atom a unique name
2197 if (AtomCount != 0) {
2198 i=0;
2199 NoNonHydrogen = 0;
2200 Walker = start;
2201 while (Walker->next != end) {
2202 Walker = Walker->next;
2203 Walker->nr = i; // update number in molecule (for easier referencing in FragmentMolecule lateron)
2204 if (Walker->type->Z != 1) // count non-hydrogen atoms whilst at it
2205 NoNonHydrogen++;
2206 Free((void **)&Walker->Name, "molecule::CountAtoms: *walker->Name");
2207 Walker->Name = (char *) Malloc(sizeof(char)*6, "molecule::CountAtoms: *walker->Name");
2208 sprintf(Walker->Name, "%2s%02d", Walker->type->symbol, Walker->nr+1);
2209 *out << "Naming atom nr. " << Walker->nr << " " << Walker->Name << "." << endl;
2210 i++;
2211 }
2212 } else
2213 *out << Verbose(3) << "AtomCount is still " << AtomCount << ", thus counting nothing." << endl;
2214 }
2215};
2216
2217/** Brings molecule::ElementCount and molecule::ElementsInMolecule up-to-date.
2218 */
2219void molecule::CountElements()
2220{
2221 int i = 0;
2222 for(i=MAX_ELEMENTS;i--;)
2223 ElementsInMolecule[i] = 0;
2224 ElementCount = 0;
2225
2226 atom *walker = start;
2227 while (walker->next != end) {
2228 walker = walker->next;
2229 ElementsInMolecule[walker->type->Z]++;
2230 i++;
2231 }
2232 for(i=MAX_ELEMENTS;i--;)
2233 ElementCount += (ElementsInMolecule[i] != 0 ? 1 : 0);
2234};
2235
2236/** Counts all cyclic bonds and returns their number.
2237 * \note Hydrogen bonds can never by cyclic, thus no check for that
2238 * \param *out output stream for debugging
2239 * \return number opf cyclic bonds
2240 */
2241int molecule::CountCyclicBonds(ofstream *out)
2242{
2243 int No = 0;
2244 int *MinimumRingSize = NULL;
2245 MoleculeLeafClass *Subgraphs = NULL;
2246 bond *Binder = first;
2247 if ((Binder->next != last) && (Binder->next->Type == Undetermined)) {
2248 *out << Verbose(0) << "No Depth-First-Search analysis performed so far, calling ..." << endl;
2249 Subgraphs = DepthFirstSearchAnalysis(out, MinimumRingSize);
2250 while (Subgraphs->next != NULL) {
2251 Subgraphs = Subgraphs->next;
2252 delete(Subgraphs->previous);
2253 }
2254 delete(Subgraphs);
2255 delete[](MinimumRingSize);
2256 }
2257 while(Binder->next != last) {
2258 Binder = Binder->next;
2259 if (Binder->Cyclic)
2260 No++;
2261 }
2262 return No;
2263};
2264/** Returns Shading as a char string.
2265 * \param color the Shading
2266 * \return string of the flag
2267 */
2268string molecule::GetColor(enum Shading color)
2269{
2270 switch(color) {
2271 case white:
2272 return "white";
2273 break;
2274 case lightgray:
2275 return "lightgray";
2276 break;
2277 case darkgray:
2278 return "darkgray";
2279 break;
2280 case black:
2281 return "black";
2282 break;
2283 default:
2284 return "uncolored";
2285 break;
2286 };
2287};
2288
2289
2290/** Counts necessary number of valence electrons and returns number and SpinType.
2291 * \param configuration containing everything
2292 */
2293void molecule::CalculateOrbitals(class config &configuration)
2294{
2295 configuration.MaxPsiDouble = configuration.PsiMaxNoDown = configuration.PsiMaxNoUp = configuration.PsiType = 0;
2296 for(int i=MAX_ELEMENTS;i--;) {
2297 if (ElementsInMolecule[i] != 0) {
2298 //cout << "CalculateOrbitals: " << elemente->FindElement(i)->name << " has a valence of " << (int)elemente->FindElement(i)->Valence << " and there are " << ElementsInMolecule[i] << " of it." << endl;
2299 configuration.MaxPsiDouble += ElementsInMolecule[i]*((int)elemente->FindElement(i)->Valence);
2300 }
2301 }
2302 configuration.PsiMaxNoDown = configuration.MaxPsiDouble/2 + (configuration.MaxPsiDouble % 2);
2303 configuration.PsiMaxNoUp = configuration.MaxPsiDouble/2;
2304 configuration.MaxPsiDouble /= 2;
2305 configuration.PsiType = (configuration.PsiMaxNoDown == configuration.PsiMaxNoUp) ? 0 : 1;
2306 if ((configuration.PsiType == 1) && (configuration.ProcPEPsi < 2)) {
2307 configuration.ProcPEGamma /= 2;
2308 configuration.ProcPEPsi *= 2;
2309 } else {
2310 configuration.ProcPEGamma *= configuration.ProcPEPsi;
2311 configuration.ProcPEPsi = 1;
2312 }
2313 configuration.InitMaxMinStopStep = configuration.MaxMinStopStep = configuration.MaxPsiDouble;
2314};
2315
2316/** Creates an adjacency list of the molecule.
2317 * Generally, we use the CSD approach to bond recognition, that is the the distance
2318 * between two atoms A and B must be within [Rcov(A)+Rcov(B)-t,Rcov(A)+Rcov(B)+t] with
2319 * a threshold t = 0.4 Angstroem.
2320 * To make it O(N log N) the function uses the linked-cell technique as follows:
2321 * The procedure is step-wise:
2322 * -# Remove every bond in list
2323 * -# Count the atoms in the molecule with CountAtoms()
2324 * -# partition cell into smaller linked cells of size \a bonddistance
2325 * -# put each atom into its corresponding cell
2326 * -# go through every cell, check the atoms therein against all possible bond partners in the 27 adjacent cells, add bond if true
2327 * -# create the list of bonds via CreateListOfBondsPerAtom()
2328 * -# correct the bond degree iteratively (single->double->triple bond)
2329 * -# finally print the bond list to \a *out if desired
2330 * \param *out out stream for printing the matrix, NULL if no output
2331 * \param bonddistance length of linked cells (i.e. maximum minimal length checked)
2332 * \param IsAngstroem whether coordinate system is gauged to Angstroem or Bohr radii
2333 */
2334void molecule::CreateAdjacencyList(ofstream *out, double bonddistance, bool IsAngstroem)
2335{
2336 atom *Walker = NULL, *OtherWalker = NULL, *Candidate = NULL;
2337 int No, NoBonds, CandidateBondNo;
2338 int NumberCells, divisor[NDIM], n[NDIM], N[NDIM], index, Index, j;
2339 molecule **CellList;
2340 double distance, MinDistance, MaxDistance;
2341 double *matrix = ReturnFullMatrixforSymmetric(cell_size);
2342 Vector x;
2343
2344 BondDistance = bonddistance; // * ((IsAngstroem) ? 1. : 1./AtomicLengthToAngstroem);
2345 *out << Verbose(0) << "Begin of CreateAdjacencyList." << endl;
2346 // remove every bond from the list
2347 if ((first->next != last) && (last->previous != first)) { // there are bonds present
2348 cleanup(first,last);
2349 }
2350
2351 // count atoms in molecule = dimension of matrix (also give each unique name and continuous numbering)
2352 CountAtoms(out);
2353 *out << Verbose(1) << "AtomCount " << AtomCount << "." << endl;
2354
2355 if (AtomCount != 0) {
2356 // 1. find divisor for each axis, such that a sphere with radius of at least bonddistance can be placed into each cell
2357 j=-1;
2358 for (int i=0;i<NDIM;i++) {
2359 j += i+1;
2360 divisor[i] = (int)floor(cell_size[j]/bonddistance); // take smaller value such that size of linked cell is at least bonddistance
2361 *out << Verbose(1) << "divisor[" << i << "] = " << divisor[i] << "." << endl;
2362 }
2363 // 2a. allocate memory for the cell list
2364 NumberCells = divisor[0]*divisor[1]*divisor[2];
2365 *out << Verbose(1) << "Allocating " << NumberCells << " cells." << endl;
2366 CellList = (molecule **) Malloc(sizeof(molecule *)*NumberCells, "molecule::CreateAdjacencyList - ** CellList");
2367 for (int i=NumberCells;i--;)
2368 CellList[i] = NULL;
2369
2370 // 2b. put all atoms into its corresponding list
2371 Walker = start;
2372 while(Walker->next != end) {
2373 Walker = Walker->next;
2374 //*out << Verbose(1) << "Current atom is " << *Walker << " with coordinates ";
2375 //Walker->x.Output(out);
2376 //*out << "." << endl;
2377 // compute the cell by the atom's coordinates
2378 j=-1;
2379 for (int i=0;i<NDIM;i++) {
2380 j += i+1;
2381 x.CopyVector(&(Walker->x));
2382 x.KeepPeriodic(out, matrix);
2383 n[i] = (int)floor(x.x[i]/cell_size[j]*(double)divisor[i]);
2384 }
2385 index = n[2] + (n[1] + n[0] * divisor[1]) * divisor[2];
2386 *out << Verbose(1) << "Atom " << *Walker << " goes into cell number [" << n[0] << "," << n[1] << "," << n[2] << "] = " << index << "." << endl;
2387 // add copy atom to this cell
2388 if (CellList[index] == NULL) // allocate molecule if not done
2389 CellList[index] = new molecule(elemente);
2390 OtherWalker = CellList[index]->AddCopyAtom(Walker); // add a copy of walker to this atom, father will be walker for later reference
2391 //*out << Verbose(1) << "Copy Atom is " << *OtherWalker << "." << endl;
2392 }
2393 //for (int i=0;i<NumberCells;i++)
2394 //*out << Verbose(1) << "Cell number " << i << ": " << CellList[i] << "." << endl;
2395
2396 // 3a. go through every cell
2397 for (N[0]=divisor[0];N[0]--;)
2398 for (N[1]=divisor[1];N[1]--;)
2399 for (N[2]=divisor[2];N[2]--;) {
2400 Index = N[2] + (N[1] + N[0] * divisor[1]) * divisor[2];
2401 if (CellList[Index] != NULL) { // if there atoms in this cell
2402 //*out << Verbose(1) << "Current cell is " << Index << "." << endl;
2403 // 3b. for every atom therein
2404 Walker = CellList[Index]->start;
2405 while (Walker->next != CellList[Index]->end) { // go through every atom
2406 Walker = Walker->next;
2407 //*out << Verbose(0) << "Current Atom is " << *Walker << "." << endl;
2408 // 3c. check for possible bond between each atom in this and every one in the 27 cells
2409 for (n[0]=-1;n[0]<=1;n[0]++)
2410 for (n[1]=-1;n[1]<=1;n[1]++)
2411 for (n[2]=-1;n[2]<=1;n[2]++) {
2412 // compute the index of this comparison cell and make it periodic
2413 index = ((N[2]+n[2]+divisor[2])%divisor[2]) + (((N[1]+n[1]+divisor[1])%divisor[1]) + ((N[0]+n[0]+divisor[0])%divisor[0]) * divisor[1]) * divisor[2];
2414 //*out << Verbose(1) << "Number of comparison cell is " << index << "." << endl;
2415 if (CellList[index] != NULL) { // if there are any atoms in this cell
2416 OtherWalker = CellList[index]->start;
2417 while(OtherWalker->next != CellList[index]->end) { // go through every atom in this cell
2418 OtherWalker = OtherWalker->next;
2419 //*out << Verbose(0) << "Current comparison atom is " << *OtherWalker << "." << endl;
2420 /// \todo periodic check is missing here!
2421 //*out << Verbose(1) << "Checking distance " << OtherWalker->x.PeriodicDistance(&(Walker->x), cell_size) << " against typical bond length of " << bonddistance*bonddistance << "." << endl;
2422 MinDistance = OtherWalker->type->CovalentRadius + Walker->type->CovalentRadius;
2423 MinDistance *= (IsAngstroem) ? 1. : 1./AtomicLengthToAngstroem;
2424 MaxDistance = MinDistance + BONDTHRESHOLD;
2425 MinDistance -= BONDTHRESHOLD;
2426 distance = OtherWalker->x.PeriodicDistance(&(Walker->x), cell_size);
2427 if ((OtherWalker->father->nr > Walker->father->nr) && (distance <= MaxDistance*MaxDistance) && (distance >= MinDistance*MinDistance)) { // create bond if distance is smaller
2428 *out << Verbose(0) << "Adding Bond between " << *Walker << " and " << *OtherWalker << "." << endl;
2429 AddBond(Walker->father, OtherWalker->father, 1); // also increases molecule::BondCount
2430 BondCount++;
2431 } else {
2432 //*out << Verbose(1) << "Not Adding: Wrong label order or distance too great." << endl;
2433 }
2434 }
2435 }
2436 }
2437 }
2438 }
2439 }
2440 // 4. free the cell again
2441 for (int i=NumberCells;i--;)
2442 if (CellList[i] != NULL) {
2443 delete(CellList[i]);
2444 }
2445 Free((void **)&CellList, "molecule::CreateAdjacencyList - ** CellList");
2446
2447 // create the adjacency list per atom
2448 CreateListOfBondsPerAtom(out);
2449
2450 // correct Bond degree of each bond by checking both bond partners for a mismatch between valence and current sum of bond degrees,
2451 // iteratively increase the one first where the other bond partner has the fewest number of bonds (i.e. in general bonds oxygene
2452 // preferred over carbon bonds). Beforehand, we had picked the first mismatching partner, which lead to oxygenes with single instead of
2453 // double bonds as was expected.
2454 if (BondCount != 0) {
2455 NoCyclicBonds = 0;
2456 *out << Verbose(1) << "Correcting Bond degree of each bond ... ";
2457 do {
2458 No = 0; // No acts as breakup flag (if 1 we still continue)
2459 Walker = start;
2460 while (Walker->next != end) { // go through every atom
2461 Walker = Walker->next;
2462 // count valence of first partner
2463 NoBonds = 0;
2464 for(j=0;j<NumberOfBondsPerAtom[Walker->nr];j++)
2465 NoBonds += ListOfBondsPerAtom[Walker->nr][j]->BondDegree;
2466 *out << Verbose(3) << "Walker " << *Walker << ": " << (int)Walker->type->NoValenceOrbitals << " > " << NoBonds << "?" << endl;
2467 if ((int)(Walker->type->NoValenceOrbitals) > NoBonds) { // we have a mismatch, check all bonding partners for mismatch
2468 Candidate = NULL;
2469 for(int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++) { // go through each of its bond partners
2470 OtherWalker = ListOfBondsPerAtom[Walker->nr][i]->GetOtherAtom(Walker);
2471 // count valence of second partner
2472 NoBonds = 0;
2473 for(j=0;j<NumberOfBondsPerAtom[OtherWalker->nr];j++)
2474 NoBonds += ListOfBondsPerAtom[OtherWalker->nr][j]->BondDegree;
2475 *out << Verbose(3) << "OtherWalker " << *OtherWalker << ": " << (int)OtherWalker->type->NoValenceOrbitals << " > " << NoBonds << "?" << endl;
2476 if ((int)(OtherWalker->type->NoValenceOrbitals) > NoBonds) { // check if possible candidate
2477 if ((Candidate == NULL) || (NumberOfBondsPerAtom[Candidate->nr] > NumberOfBondsPerAtom[OtherWalker->nr])) { // pick the one with fewer number of bonds first
2478 Candidate = OtherWalker;
2479 CandidateBondNo = i;
2480 *out << Verbose(3) << "New candidate is " << *Candidate << "." << endl;
2481 }
2482 }
2483 }
2484 if (Candidate != NULL) {
2485 ListOfBondsPerAtom[Walker->nr][CandidateBondNo]->BondDegree++;
2486 *out << Verbose(2) << "Increased bond degree for bond " << *ListOfBondsPerAtom[Walker->nr][CandidateBondNo] << "." << endl;
2487 }
2488 }
2489 }
2490 } while (No);
2491 *out << " done." << endl;
2492 } else
2493 *out << Verbose(1) << "BondCount is " << BondCount << ", no bonds between any of the " << AtomCount << " atoms." << endl;
2494 *out << Verbose(1) << "I detected " << BondCount << " bonds in the molecule with distance " << bonddistance << "." << endl;
2495
2496 // output bonds for debugging (if bond chain list was correctly installed)
2497 *out << Verbose(1) << endl << "From contents of bond chain list:";
2498 bond *Binder = first;
2499 while(Binder->next != last) {
2500 Binder = Binder->next;
2501 *out << *Binder << "\t" << endl;
2502 }
2503 *out << endl;
2504 } else
2505 *out << Verbose(1) << "AtomCount is " << AtomCount << ", thus no bonds, no connections!." << endl;
2506 *out << Verbose(0) << "End of CreateAdjacencyList." << endl;
2507 Free((void **)&matrix, "molecule::CreateAdjacencyList: *matrix");
2508};
2509
2510/** Performs a Depth-First search on this molecule.
2511 * Marks bonds in molecule as cyclic, bridge, ... and atoms as
2512 * articulations points, ...
2513 * We use the algorithm from [Even, Graph Algorithms, p.62].
2514 * \param *out output stream for debugging
2515 * \param *&MinimumRingSize contains smallest ring size in molecular structure on return or -1 if no rings were found
2516 * \return list of each disconnected subgraph as an individual molecule class structure
2517 */
2518MoleculeLeafClass * molecule::DepthFirstSearchAnalysis(ofstream *out, int *&MinimumRingSize)
2519{
2520 class StackClass<atom *> *AtomStack;
2521 AtomStack = new StackClass<atom *>(AtomCount);
2522 class StackClass<bond *> *BackEdgeStack = new StackClass<bond *> (BondCount);
2523 MoleculeLeafClass *SubGraphs = new MoleculeLeafClass(NULL);
2524 MoleculeLeafClass *LeafWalker = SubGraphs;
2525 int CurrentGraphNr = 0, OldGraphNr;
2526 int ComponentNumber = 0;
2527 atom *Walker = NULL, *OtherAtom = NULL, *Root = start->next;
2528 bond *Binder = NULL;
2529 bool BackStepping = false;
2530
2531 *out << Verbose(0) << "Begin of DepthFirstSearchAnalysis" << endl;
2532
2533 ResetAllBondsToUnused();
2534 ResetAllAtomNumbers();
2535 InitComponentNumbers();
2536 BackEdgeStack->ClearStack();
2537 while (Root != end) { // if there any atoms at all
2538 // (1) mark all edges unused, empty stack, set atom->GraphNr = 0 for all
2539 AtomStack->ClearStack();
2540
2541 // put into new subgraph molecule and add this to list of subgraphs
2542 LeafWalker = new MoleculeLeafClass(LeafWalker);
2543 LeafWalker->Leaf = new molecule(elemente);
2544 LeafWalker->Leaf->AddCopyAtom(Root);
2545
2546 OldGraphNr = CurrentGraphNr;
2547 Walker = Root;
2548 do { // (10)
2549 do { // (2) set number and Lowpoint of Atom to i, increase i, push current atom
2550 if (!BackStepping) { // if we don't just return from (8)
2551 Walker->GraphNr = CurrentGraphNr;
2552 Walker->LowpointNr = CurrentGraphNr;
2553 *out << Verbose(1) << "Setting Walker[" << Walker->Name << "]'s number to " << Walker->GraphNr << " with Lowpoint " << Walker->LowpointNr << "." << endl;
2554 AtomStack->Push(Walker);
2555 CurrentGraphNr++;
2556 }
2557 do { // (3) if Walker has no unused egdes, go to (5)
2558 BackStepping = false; // reset backstepping flag for (8)
2559 if (Binder == NULL) // if we don't just return from (11), Binder is already set to next unused
2560 Binder = FindNextUnused(Walker);
2561 if (Binder == NULL)
2562 break;
2563 *out << Verbose(2) << "Current Unused Bond is " << *Binder << "." << endl;
2564 // (4) Mark Binder used, ...
2565 Binder->MarkUsed(black);
2566 OtherAtom = Binder->GetOtherAtom(Walker);
2567 *out << Verbose(2) << "(4) OtherAtom is " << OtherAtom->Name << "." << endl;
2568 if (OtherAtom->GraphNr != -1) {
2569 // (4a) ... if "other" atom has been visited (GraphNr != 0), set lowpoint to minimum of both, go to (3)
2570 Binder->Type = BackEdge;
2571 BackEdgeStack->Push(Binder);
2572 Walker->LowpointNr = ( Walker->LowpointNr < OtherAtom->GraphNr ) ? Walker->LowpointNr : OtherAtom->GraphNr;
2573 *out << Verbose(3) << "(4a) Visited: Setting Lowpoint of Walker[" << Walker->Name << "] to " << Walker->LowpointNr << "." << endl;
2574 } else {
2575 // (4b) ... otherwise set OtherAtom as Ancestor of Walker and Walker as OtherAtom, go to (2)
2576 Binder->Type = TreeEdge;
2577 OtherAtom->Ancestor = Walker;
2578 Walker = OtherAtom;
2579 *out << Verbose(3) << "(4b) Not Visited: OtherAtom[" << OtherAtom->Name << "]'s Ancestor is now " << OtherAtom->Ancestor->Name << ", Walker is OtherAtom " << OtherAtom->Name << "." << endl;
2580 break;
2581 }
2582 Binder = NULL;
2583 } while (1); // (3)
2584 if (Binder == NULL) {
2585 *out << Verbose(2) << "No more Unused Bonds." << endl;
2586 break;
2587 } else
2588 Binder = NULL;
2589 } while (1); // (2)
2590
2591 // if we came from backstepping, yet there were no more unused bonds, we end up here with no Ancestor, because Walker is Root! Then we are finished!
2592 if ((Walker == Root) && (Binder == NULL))
2593 break;
2594
2595 // (5) if Ancestor of Walker is ...
2596 *out << Verbose(1) << "(5) Number of Walker[" << Walker->Name << "]'s Ancestor[" << Walker->Ancestor->Name << "] is " << Walker->Ancestor->GraphNr << "." << endl;
2597 if (Walker->Ancestor->GraphNr != Root->GraphNr) {
2598 // (6) (Ancestor of Walker is not Root)
2599 if (Walker->LowpointNr < Walker->Ancestor->GraphNr) {
2600 // (6a) set Ancestor's Lowpoint number to minimum of of its Ancestor and itself, go to Step(8)
2601 Walker->Ancestor->LowpointNr = (Walker->Ancestor->LowpointNr < Walker->LowpointNr) ? Walker->Ancestor->LowpointNr : Walker->LowpointNr;
2602 *out << Verbose(2) << "(6) Setting Walker[" << Walker->Name << "]'s Ancestor[" << Walker->Ancestor->Name << "]'s Lowpoint to " << Walker->Ancestor->LowpointNr << "." << endl;
2603 } else {
2604 // (7) (Ancestor of Walker is a separating vertex, remove all from stack till Walker (including), these and Ancestor form a component
2605 Walker->Ancestor->SeparationVertex = true;
2606 *out << Verbose(2) << "(7) Walker[" << Walker->Name << "]'s Ancestor[" << Walker->Ancestor->Name << "]'s is a separating vertex, creating component." << endl;
2607 SetNextComponentNumber(Walker->Ancestor, ComponentNumber);
2608 *out << Verbose(3) << "(7) Walker[" << Walker->Name << "]'s Ancestor's Compont is " << ComponentNumber << "." << endl;
2609 SetNextComponentNumber(Walker, ComponentNumber);
2610 *out << Verbose(3) << "(7) Walker[" << Walker->Name << "]'s Compont is " << ComponentNumber << "." << endl;
2611 do {
2612 OtherAtom = AtomStack->PopLast();
2613 LeafWalker->Leaf->AddCopyAtom(OtherAtom);
2614 SetNextComponentNumber(OtherAtom, ComponentNumber);
2615 *out << Verbose(3) << "(7) Other[" << OtherAtom->Name << "]'s Compont is " << ComponentNumber << "." << endl;
2616 } while (OtherAtom != Walker);
2617 ComponentNumber++;
2618 }
2619 // (8) Walker becomes its Ancestor, go to (3)
2620 *out << Verbose(2) << "(8) Walker[" << Walker->Name << "] is now its Ancestor " << Walker->Ancestor->Name << ", backstepping. " << endl;
2621 Walker = Walker->Ancestor;
2622 BackStepping = true;
2623 }
2624 if (!BackStepping) { // coming from (8) want to go to (3)
2625 // (9) remove all from stack till Walker (including), these and Root form a component
2626 AtomStack->Output(out);
2627 SetNextComponentNumber(Root, ComponentNumber);
2628 *out << Verbose(3) << "(9) Root[" << Root->Name << "]'s Component is " << ComponentNumber << "." << endl;
2629 SetNextComponentNumber(Walker, ComponentNumber);
2630 *out << Verbose(3) << "(9) Walker[" << Walker->Name << "]'s Component is " << ComponentNumber << "." << endl;
2631 do {
2632 OtherAtom = AtomStack->PopLast();
2633 LeafWalker->Leaf->AddCopyAtom(OtherAtom);
2634 SetNextComponentNumber(OtherAtom, ComponentNumber);
2635 *out << Verbose(3) << "(7) Other[" << OtherAtom->Name << "]'s Compont is " << ComponentNumber << "." << endl;
2636 } while (OtherAtom != Walker);
2637 ComponentNumber++;
2638
2639 // (11) Root is separation vertex, set Walker to Root and go to (4)
2640 Walker = Root;
2641 Binder = FindNextUnused(Walker);
2642 *out << Verbose(1) << "(10) Walker is Root[" << Root->Name << "], next Unused Bond is " << Binder << "." << endl;
2643 if (Binder != NULL) { // Root is separation vertex
2644 *out << Verbose(1) << "(11) Root is a separation vertex." << endl;
2645 Walker->SeparationVertex = true;
2646 }
2647 }
2648 } while ((BackStepping) || (Binder != NULL)); // (10) halt only if Root has no unused edges
2649
2650 // From OldGraphNr to CurrentGraphNr ranges an disconnected subgraph
2651 *out << Verbose(0) << "Disconnected subgraph ranges from " << OldGraphNr << " to " << CurrentGraphNr << "." << endl;
2652 LeafWalker->Leaf->Output(out);
2653 *out << endl;
2654
2655 // step on to next root
2656 while ((Root != end) && (Root->GraphNr != -1)) {
2657 //*out << Verbose(1) << "Current next subgraph root candidate is " << Root->Name << "." << endl;
2658 if (Root->GraphNr != -1) // if already discovered, step on
2659 Root = Root->next;
2660 }
2661 }
2662 // set cyclic bond criterium on "same LP" basis
2663 Binder = first;
2664 while(Binder->next != last) {
2665 Binder = Binder->next;
2666 if (Binder->rightatom->LowpointNr == Binder->leftatom->LowpointNr) { // cyclic ??
2667 Binder->Cyclic = true;
2668 NoCyclicBonds++;
2669 }
2670 }
2671
2672 // analysis of the cycles (print rings, get minimum cycle length)
2673 CyclicStructureAnalysis(out, BackEdgeStack, MinimumRingSize);
2674
2675 *out << Verbose(1) << "Final graph info for each atom is:" << endl;
2676 Walker = start;
2677 while (Walker->next != end) {
2678 Walker = Walker->next;
2679 *out << Verbose(2) << "Atom " << Walker->Name << " is " << ((Walker->SeparationVertex) ? "a" : "not a") << " separation vertex, components are ";
2680 OutputComponentNumber(out, Walker);
2681 *out << " with Lowpoint " << Walker->LowpointNr << " and Graph Nr. " << Walker->GraphNr << "." << endl;
2682 }
2683
2684 *out << Verbose(1) << "Final graph info for each bond is:" << endl;
2685 Binder = first;
2686 while(Binder->next != last) {
2687 Binder = Binder->next;
2688 *out << Verbose(2) << ((Binder->Type == TreeEdge) ? "TreeEdge " : "BackEdge ") << *Binder << ": <";
2689 *out << ((Binder->leftatom->SeparationVertex) ? "SP," : "") << "L" << Binder->leftatom->LowpointNr << " G" << Binder->leftatom->GraphNr << " Comp.";
2690 OutputComponentNumber(out, Binder->leftatom);
2691 *out << " === ";
2692 *out << ((Binder->rightatom->SeparationVertex) ? "SP," : "") << "L" << Binder->rightatom->LowpointNr << " G" << Binder->rightatom->GraphNr << " Comp.";
2693 OutputComponentNumber(out, Binder->rightatom);
2694 *out << ">." << endl;
2695 if (Binder->Cyclic) // cyclic ??
2696 *out << Verbose(3) << "Lowpoint at each side are equal: CYCLIC!" << endl;
2697 }
2698
2699 // free all and exit
2700 delete(AtomStack);
2701 *out << Verbose(0) << "End of DepthFirstSearchAnalysis" << endl;
2702 return SubGraphs;
2703};
2704
2705/** Analyses the cycles found and returns minimum of all cycle lengths.
2706 * We begin with a list of Back edges found during DepthFirstSearchAnalysis(). We go through this list - one end is the Root,
2707 * the other our initial Walker - and do a Breadth First Search for the Root. We mark down each Predecessor and as soon as
2708 * we have found the Root via BFS, we may climb back the closed cycle via the Predecessors. Thereby we mark atoms and bonds
2709 * as cyclic and print out the cycles.
2710 * \param *out output stream for debugging
2711 * \param *BackEdgeStack stack with all back edges found during DFS scan
2712 * \param *&MinimumRingSize contains smallest ring size in molecular structure on return or -1 if no rings were found, if set is maximum search distance
2713 * \todo BFS from the not-same-LP to find back to starting point of tributary cycle over more than one bond
2714 */
2715void molecule::CyclicStructureAnalysis(ofstream *out, class StackClass<bond *> * BackEdgeStack, int *&MinimumRingSize)
2716{
2717 atom **PredecessorList = (atom **) Malloc(sizeof(atom *)*AtomCount, "molecule::CyclicStructureAnalysis: **PredecessorList");
2718 int *ShortestPathList = (int *) Malloc(sizeof(int)*AtomCount, "molecule::CyclicStructureAnalysis: *ShortestPathList");
2719 enum Shading *ColorList = (enum Shading *) Malloc(sizeof(enum Shading)*AtomCount, "molecule::CyclicStructureAnalysis: *ColorList");
2720 class StackClass<atom *> *BFSStack = new StackClass<atom *> (AtomCount); // will hold the current ring
2721 class StackClass<atom *> *TouchedStack = new StackClass<atom *> (AtomCount); // contains all "touched" atoms (that need to be reset after BFS loop)
2722 atom *Walker = NULL, *OtherAtom = NULL, *Root = NULL;
2723 bond *Binder = NULL, *BackEdge = NULL;
2724 int RingSize, NumCycles, MinRingSize = -1;
2725
2726 // initialise each vertex as white with no predecessor, empty queue, color Root lightgray
2727 for (int i=AtomCount;i--;) {
2728 PredecessorList[i] = NULL;
2729 ShortestPathList[i] = -1;
2730 ColorList[i] = white;
2731 }
2732 MinimumRingSize = new int[AtomCount];
2733 for(int i=AtomCount;i--;)
2734 MinimumRingSize[i] = AtomCount;
2735
2736
2737 *out << Verbose(1) << "Back edge list - ";
2738 BackEdgeStack->Output(out);
2739
2740 *out << Verbose(1) << "Analysing cycles ... " << endl;
2741 NumCycles = 0;
2742 while (!BackEdgeStack->IsEmpty()) {
2743 BackEdge = BackEdgeStack->PopFirst();
2744 // this is the target
2745 Root = BackEdge->leftatom;
2746 // this is the source point
2747 Walker = BackEdge->rightatom;
2748 ShortestPathList[Walker->nr] = 0;
2749 BFSStack->ClearStack(); // start with empty BFS stack
2750 BFSStack->Push(Walker);
2751 TouchedStack->Push(Walker);
2752 *out << Verbose(1) << "---------------------------------------------------------------------------------------------------------" << endl;
2753 OtherAtom = NULL;
2754 do { // look for Root
2755 Walker = BFSStack->PopFirst();
2756 *out << Verbose(2) << "Current Walker is " << *Walker << ", we look for SP to Root " << *Root << "." << endl;
2757 for(int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++) {
2758 Binder = ListOfBondsPerAtom[Walker->nr][i];
2759 if (Binder != BackEdge) { // only walk along DFS spanning tree (otherwise we always find SP of one being backedge Binder)
2760 OtherAtom = Binder->GetOtherAtom(Walker);
2761#ifdef ADDHYDROGEN
2762 if (OtherAtom->type->Z != 1) {
2763#endif
2764 *out << Verbose(2) << "Current OtherAtom is: " << OtherAtom->Name << " for bond " << *Binder << "." << endl;
2765 if (ColorList[OtherAtom->nr] == white) {
2766 TouchedStack->Push(OtherAtom);
2767 ColorList[OtherAtom->nr] = lightgray;
2768 PredecessorList[OtherAtom->nr] = Walker; // Walker is the predecessor
2769 ShortestPathList[OtherAtom->nr] = ShortestPathList[Walker->nr]+1;
2770 *out << Verbose(2) << "Coloring OtherAtom " << OtherAtom->Name << " lightgray, its predecessor is " << Walker->Name << " and its Shortest Path is " << ShortestPathList[OtherAtom->nr] << " egde(s) long." << endl;
2771 //if (ShortestPathList[OtherAtom->nr] < MinimumRingSize[Walker->GetTrueFather()->nr]) { // Check for maximum distance
2772 *out << Verbose(3) << "Putting OtherAtom into queue." << endl;
2773 BFSStack->Push(OtherAtom);
2774 //}
2775 } else {
2776 *out << Verbose(3) << "Not Adding, has already been visited." << endl;
2777 }
2778 if (OtherAtom == Root)
2779 break;
2780#ifdef ADDHYDROGEN
2781 } else {
2782 *out << Verbose(2) << "Skipping hydrogen atom " << *OtherAtom << "." << endl;
2783 ColorList[OtherAtom->nr] = black;
2784 }
2785#endif
2786 } else {
2787 *out << Verbose(2) << "Bond " << *Binder << " not Visiting, is the back edge." << endl;
2788 }
2789 }
2790 ColorList[Walker->nr] = black;
2791 *out << Verbose(1) << "Coloring Walker " << Walker->Name << " black." << endl;
2792 if (OtherAtom == Root) { // if we have found the root, check whether this cycle wasn't already found beforehand
2793 // step through predecessor list
2794 while (OtherAtom != BackEdge->rightatom) {
2795 if (!OtherAtom->GetTrueFather()->IsCyclic) // if one bond in the loop is not marked as cyclic, we haven't found this cycle yet
2796 break;
2797 else
2798 OtherAtom = PredecessorList[OtherAtom->nr];
2799 }
2800 if (OtherAtom == BackEdge->rightatom) { // if each atom in found cycle is cyclic, loop's been found before already
2801 *out << Verbose(3) << "This cycle was already found before, skipping and removing seeker from search." << endl;\
2802 do {
2803 OtherAtom = TouchedStack->PopLast();
2804 if (PredecessorList[OtherAtom->nr] == Walker) {
2805 *out << Verbose(4) << "Removing " << *OtherAtom << " from lists and stacks." << endl;
2806 PredecessorList[OtherAtom->nr] = NULL;
2807 ShortestPathList[OtherAtom->nr] = -1;
2808 ColorList[OtherAtom->nr] = white;
2809 BFSStack->RemoveItem(OtherAtom);
2810 }
2811 } while ((!TouchedStack->IsEmpty()) && (PredecessorList[OtherAtom->nr] == NULL));
2812 TouchedStack->Push(OtherAtom); // last was wrongly popped
2813 OtherAtom = BackEdge->rightatom; // set to not Root
2814 } else
2815 OtherAtom = Root;
2816 }
2817 } while ((!BFSStack->IsEmpty()) && (OtherAtom != Root) && (OtherAtom != NULL)); // || (ShortestPathList[OtherAtom->nr] < MinimumRingSize[Walker->GetTrueFather()->nr])));
2818
2819 if (OtherAtom == Root) {
2820 // now climb back the predecessor list and thus find the cycle members
2821 NumCycles++;
2822 RingSize = 1;
2823 Root->GetTrueFather()->IsCyclic = true;
2824 *out << Verbose(1) << "Found ring contains: ";
2825 Walker = Root;
2826 while (Walker != BackEdge->rightatom) {
2827 *out << Walker->Name << " <-> ";
2828 Walker = PredecessorList[Walker->nr];
2829 Walker->GetTrueFather()->IsCyclic = true;
2830 RingSize++;
2831 }
2832 *out << Walker->Name << " with a length of " << RingSize << "." << endl << endl;
2833 // walk through all and set MinimumRingSize
2834 Walker = Root;
2835 MinimumRingSize[Walker->GetTrueFather()->nr] = RingSize;
2836 while (Walker != BackEdge->rightatom) {
2837 Walker = PredecessorList[Walker->nr];
2838 if (RingSize < MinimumRingSize[Walker->GetTrueFather()->nr])
2839 MinimumRingSize[Walker->GetTrueFather()->nr] = RingSize;
2840 }
2841 if ((RingSize < MinRingSize) || (MinRingSize == -1))
2842 MinRingSize = RingSize;
2843 } else {
2844 *out << Verbose(1) << "No ring containing " << *Root << " with length equal to or smaller than " << MinimumRingSize[Walker->GetTrueFather()->nr] << " found." << endl;
2845 }
2846
2847 // now clean the lists
2848 while (!TouchedStack->IsEmpty()){
2849 Walker = TouchedStack->PopFirst();
2850 PredecessorList[Walker->nr] = NULL;
2851 ShortestPathList[Walker->nr] = -1;
2852 ColorList[Walker->nr] = white;
2853 }
2854 }
2855 if (MinRingSize != -1) {
2856 // go over all atoms
2857 Root = start;
2858 while(Root->next != end) {
2859 Root = Root->next;
2860
2861 if (MinimumRingSize[Root->GetTrueFather()->nr] == AtomCount) { // check whether MinimumRingSize is set, if not BFS to next where it is
2862 Walker = Root;
2863 ShortestPathList[Walker->nr] = 0;
2864 BFSStack->ClearStack(); // start with empty BFS stack
2865 BFSStack->Push(Walker);
2866 TouchedStack->Push(Walker);
2867 //*out << Verbose(1) << "---------------------------------------------------------------------------------------------------------" << endl;
2868 OtherAtom = Walker;
2869 while (OtherAtom != NULL) { // look for Root
2870 Walker = BFSStack->PopFirst();
2871 //*out << Verbose(2) << "Current Walker is " << *Walker << ", we look for SP to Root " << *Root << "." << endl;
2872 for(int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++) {
2873 Binder = ListOfBondsPerAtom[Walker->nr][i];
2874 if ((Binder != BackEdge) || (NumberOfBondsPerAtom[Walker->nr] == 1)) { // only walk along DFS spanning tree (otherwise we always find SP of 1 being backedge Binder), but terminal hydrogens may be connected via backedge, hence extra check
2875 OtherAtom = Binder->GetOtherAtom(Walker);
2876 //*out << Verbose(2) << "Current OtherAtom is: " << OtherAtom->Name << " for bond " << *Binder << "." << endl;
2877 if (ColorList[OtherAtom->nr] == white) {
2878 TouchedStack->Push(OtherAtom);
2879 ColorList[OtherAtom->nr] = lightgray;
2880 PredecessorList[OtherAtom->nr] = Walker; // Walker is the predecessor
2881 ShortestPathList[OtherAtom->nr] = ShortestPathList[Walker->nr]+1;
2882 //*out << Verbose(2) << "Coloring OtherAtom " << OtherAtom->Name << " lightgray, its predecessor is " << Walker->Name << " and its Shortest Path is " << ShortestPathList[OtherAtom->nr] << " egde(s) long." << endl;
2883 if (OtherAtom->GetTrueFather()->IsCyclic) { // if the other atom is connected to a ring
2884 MinimumRingSize[Root->GetTrueFather()->nr] = ShortestPathList[OtherAtom->nr]+MinimumRingSize[OtherAtom->GetTrueFather()->nr];
2885 OtherAtom = NULL; //break;
2886 break;
2887 } else
2888 BFSStack->Push(OtherAtom);
2889 } else {
2890 //*out << Verbose(3) << "Not Adding, has already been visited." << endl;
2891 }
2892 } else {
2893 //*out << Verbose(3) << "Not Visiting, is a back edge." << endl;
2894 }
2895 }
2896 ColorList[Walker->nr] = black;
2897 //*out << Verbose(1) << "Coloring Walker " << Walker->Name << " black." << endl;
2898 }
2899
2900 // now clean the lists
2901 while (!TouchedStack->IsEmpty()){
2902 Walker = TouchedStack->PopFirst();
2903 PredecessorList[Walker->nr] = NULL;
2904 ShortestPathList[Walker->nr] = -1;
2905 ColorList[Walker->nr] = white;
2906 }
2907 }
2908 *out << Verbose(1) << "Minimum ring size of " << *Root << " is " << MinimumRingSize[Root->GetTrueFather()->nr] << "." << endl;
2909 }
2910 *out << Verbose(1) << "Minimum ring size is " << MinRingSize << ", over " << NumCycles << " cycles total." << endl;
2911 } else
2912 *out << Verbose(1) << "No rings were detected in the molecular structure." << endl;
2913
2914 Free((void **)&PredecessorList, "molecule::CyclicStructureAnalysis: **PredecessorList");
2915 Free((void **)&ShortestPathList, "molecule::CyclicStructureAnalysis: **ShortestPathList");
2916 Free((void **)&ColorList, "molecule::CyclicStructureAnalysis: **ColorList");
2917 delete(BFSStack);
2918};
2919
2920/** Sets the next component number.
2921 * This is O(N) as the number of bonds per atom is bound.
2922 * \param *vertex atom whose next atom::*ComponentNr is to be set
2923 * \param nr number to use
2924 */
2925void molecule::SetNextComponentNumber(atom *vertex, int nr)
2926{
2927 int i=0;
2928 if (vertex != NULL) {
2929 for(;i<NumberOfBondsPerAtom[vertex->nr];i++) {
2930 if (vertex->ComponentNr[i] == -1) { // check if not yet used
2931 vertex->ComponentNr[i] = nr;
2932 break;
2933 }
2934 else if (vertex->ComponentNr[i] == nr) // if number is already present, don't add another time
2935 break; // breaking here will not cause error!
2936 }
2937 if (i == NumberOfBondsPerAtom[vertex->nr])
2938 cerr << "Error: All Component entries are already occupied!" << endl;
2939 } else
2940 cerr << "Error: Given vertex is NULL!" << endl;
2941};
2942
2943/** Output a list of flags, stating whether the bond was visited or not.
2944 * \param *out output stream for debugging
2945 */
2946void molecule::OutputComponentNumber(ofstream *out, atom *vertex)
2947{
2948 for(int i=0;i<NumberOfBondsPerAtom[vertex->nr];i++)
2949 *out << vertex->ComponentNr[i] << " ";
2950};
2951
2952/** Allocates memory for all atom::*ComponentNr in this molecule and sets each entry to -1.
2953 */
2954void molecule::InitComponentNumbers()
2955{
2956 atom *Walker = start;
2957 while(Walker->next != end) {
2958 Walker = Walker->next;
2959 if (Walker->ComponentNr != NULL)
2960 Free((void **)&Walker->ComponentNr, "molecule::InitComponentNumbers: **Walker->ComponentNr");
2961 Walker->ComponentNr = (int *) Malloc(sizeof(int)*NumberOfBondsPerAtom[Walker->nr], "molecule::InitComponentNumbers: *Walker->ComponentNr");
2962 for (int i=NumberOfBondsPerAtom[Walker->nr];i--;)
2963 Walker->ComponentNr[i] = -1;
2964 }
2965};
2966
2967/** Returns next unused bond for this atom \a *vertex or NULL of none exists.
2968 * \param *vertex atom to regard
2969 * \return bond class or NULL
2970 */
2971bond * molecule::FindNextUnused(atom *vertex)
2972{
2973 for(int i=0;i<NumberOfBondsPerAtom[vertex->nr];i++)
2974 if (ListOfBondsPerAtom[vertex->nr][i]->IsUsed() == white)
2975 return(ListOfBondsPerAtom[vertex->nr][i]);
2976 return NULL;
2977};
2978
2979/** Resets bond::Used flag of all bonds in this molecule.
2980 * \return true - success, false - -failure
2981 */
2982void molecule::ResetAllBondsToUnused()
2983{
2984 bond *Binder = first;
2985 while (Binder->next != last) {
2986 Binder = Binder->next;
2987 Binder->ResetUsed();
2988 }
2989};
2990
2991/** Resets atom::nr to -1 of all atoms in this molecule.
2992 */
2993void molecule::ResetAllAtomNumbers()
2994{
2995 atom *Walker = start;
2996 while (Walker->next != end) {
2997 Walker = Walker->next;
2998 Walker->GraphNr = -1;
2999 }
3000};
3001
3002/** Output a list of flags, stating whether the bond was visited or not.
3003 * \param *out output stream for debugging
3004 * \param *list
3005 */
3006void OutputAlreadyVisited(ofstream *out, int *list)
3007{
3008 *out << Verbose(4) << "Already Visited Bonds:\t";
3009 for(int i=1;i<=list[0];i++) *out << Verbose(0) << list[i] << " ";
3010 *out << endl;
3011};
3012
3013/** Estimates by educated guessing (using upper limit) the expected number of fragments.
3014 * The upper limit is
3015 * \f[
3016 * n = N \cdot C^k
3017 * \f]
3018 * where \f$C=2^c\f$ and c is the maximum bond degree over N number of atoms.
3019 * \param *out output stream for debugging
3020 * \param order bond order k
3021 * \return number n of fragments
3022 */
3023int molecule::GuesstimateFragmentCount(ofstream *out, int order)
3024{
3025 int c = 0;
3026 int FragmentCount;
3027 // get maximum bond degree
3028 atom *Walker = start;
3029 while (Walker->next != end) {
3030 Walker = Walker->next;
3031 c = (NumberOfBondsPerAtom[Walker->nr] > c) ? NumberOfBondsPerAtom[Walker->nr] : c;
3032 }
3033 FragmentCount = NoNonHydrogen*(1 << (c*order));
3034 *out << Verbose(1) << "Upper limit for this subgraph is " << FragmentCount << " for " << NoNonHydrogen << " non-H atoms with maximum bond degree of " << c << "." << endl;
3035 return FragmentCount;
3036};
3037
3038/** Scans a single line for number and puts them into \a KeySet.
3039 * \param *out output stream for debugging
3040 * \param *buffer buffer to scan
3041 * \param &CurrentSet filled KeySet on return
3042 * \return true - at least one valid atom id parsed, false - CurrentSet is empty
3043 */
3044bool molecule::ScanBufferIntoKeySet(ofstream *out, char *buffer, KeySet &CurrentSet)
3045{
3046 stringstream line;
3047 int AtomNr;
3048 int status = 0;
3049
3050 line.str(buffer);
3051 while (!line.eof()) {
3052 line >> AtomNr;
3053 if ((AtomNr >= 0) && (AtomNr < AtomCount)) {
3054 CurrentSet.insert(AtomNr); // insert at end, hence in same order as in file!
3055 status++;
3056 } // else it's "-1" or else and thus must not be added
3057 }
3058 *out << Verbose(1) << "The scanned KeySet is ";
3059 for(KeySet::iterator runner = CurrentSet.begin(); runner != CurrentSet.end(); runner++) {
3060 *out << (*runner) << "\t";
3061 }
3062 *out << endl;
3063 return (status != 0);
3064};
3065
3066/** Parses the KeySet file and fills \a *FragmentList from the known molecule structure.
3067 * Does two-pass scanning:
3068 * -# Scans the keyset file and initialises a temporary graph
3069 * -# Scans TEFactors file and sets the TEFactor of each key set in the temporary graph accordingly
3070 * Finally, the temporary graph is inserted into the given \a FragmentList for return.
3071 * \param *out output stream for debugging
3072 * \param *path path to file
3073 * \param *FragmentList empty, filled on return
3074 * \return true - parsing successfully, false - failure on parsing (FragmentList will be NULL)
3075 */
3076bool molecule::ParseKeySetFile(ofstream *out, char *path, Graph *&FragmentList)
3077{
3078 bool status = true;
3079 ifstream InputFile;
3080 stringstream line;
3081 GraphTestPair testGraphInsert;
3082 int NumberOfFragments = 0;
3083 double TEFactor;
3084 char *filename = (char *) Malloc(sizeof(char)*MAXSTRINGSIZE, "molecule::ParseKeySetFile - filename");
3085
3086 if (FragmentList == NULL) { // check list pointer
3087 FragmentList = new Graph;
3088 }
3089
3090 // 1st pass: open file and read
3091 *out << Verbose(1) << "Parsing the KeySet file ... " << endl;
3092 sprintf(filename, "%s/%s%s", path, FRAGMENTPREFIX, KEYSETFILE);
3093 InputFile.open(filename);
3094 if (InputFile != NULL) {
3095 // each line represents a new fragment
3096 char *buffer = (char *) Malloc(sizeof(char)*MAXSTRINGSIZE, "molecule::ParseKeySetFile - *buffer");
3097 // 1. parse keysets and insert into temp. graph
3098 while (!InputFile.eof()) {
3099 InputFile.getline(buffer, MAXSTRINGSIZE);
3100 KeySet CurrentSet;
3101 if ((strlen(buffer) > 0) && (ScanBufferIntoKeySet(out, buffer, CurrentSet))) { // if at least one valid atom was added, write config
3102 testGraphInsert = FragmentList->insert(GraphPair (CurrentSet,pair<int,double>(NumberOfFragments++,1))); // store fragment number and current factor
3103 if (!testGraphInsert.second) {
3104 cerr << "KeySet file must be corrupt as there are two equal key sets therein!" << endl;
3105 }
3106 //FragmentList->ListOfMolecules[NumberOfFragments++] = StoreFragmentFromKeySet(out, CurrentSet, IsAngstroem);
3107 }
3108 }
3109 // 2. Free and done
3110 InputFile.close();
3111 InputFile.clear();
3112 Free((void **)&buffer, "molecule::ParseKeySetFile - *buffer");
3113 *out << Verbose(1) << "done." << endl;
3114 } else {
3115 *out << Verbose(1) << "File " << filename << " not found." << endl;
3116 status = false;
3117 }
3118
3119 // 2nd pass: open TEFactors file and read
3120 *out << Verbose(1) << "Parsing the TEFactors file ... " << endl;
3121 sprintf(filename, "%s/%s%s", path, FRAGMENTPREFIX, TEFACTORSFILE);
3122 InputFile.open(filename);
3123 if (InputFile != NULL) {
3124 // 3. add found TEFactors to each keyset
3125 NumberOfFragments = 0;
3126 for(Graph::iterator runner = FragmentList->begin();runner != FragmentList->end(); runner++) {
3127 if (!InputFile.eof()) {
3128 InputFile >> TEFactor;
3129 (*runner).second.second = TEFactor;
3130 *out << Verbose(2) << "Setting " << ++NumberOfFragments << " fragment's TEFactor to " << (*runner).second.second << "." << endl;
3131 } else {
3132 status = false;
3133 break;
3134 }
3135 }
3136 // 4. Free and done
3137 InputFile.close();
3138 *out << Verbose(1) << "done." << endl;
3139 } else {
3140 *out << Verbose(1) << "File " << filename << " not found." << endl;
3141 status = false;
3142 }
3143
3144 // free memory
3145 Free((void **)&filename, "molecule::ParseKeySetFile - filename");
3146
3147 return status;
3148};
3149
3150/** Stores keysets and TEFactors to file.
3151 * \param *out output stream for debugging
3152 * \param KeySetList Graph with Keysets and factors
3153 * \param *path path to file
3154 * \return true - file written successfully, false - writing failed
3155 */
3156bool molecule::StoreKeySetFile(ofstream *out, Graph &KeySetList, char *path)
3157{
3158 ofstream output;
3159 bool status = true;
3160 string line;
3161
3162 // open KeySet file
3163 line = path;
3164 line.append("/");
3165 line += FRAGMENTPREFIX;
3166 line += KEYSETFILE;
3167 output.open(line.c_str(), ios::out);
3168 *out << Verbose(1) << "Saving key sets of the total graph ... ";
3169 if(output != NULL) {
3170 for(Graph::iterator runner = KeySetList.begin(); runner != KeySetList.end(); runner++) {
3171 for (KeySet::iterator sprinter = (*runner).first.begin();sprinter != (*runner).first.end(); sprinter++) {
3172 if (sprinter != (*runner).first.begin())
3173 output << "\t";
3174 output << *sprinter;
3175 }
3176 output << endl;
3177 }
3178 *out << "done." << endl;
3179 } else {
3180 cerr << "Unable to open " << line << " for writing keysets!" << endl;
3181 status = false;
3182 }
3183 output.close();
3184 output.clear();
3185
3186 // open TEFactors file
3187 line = path;
3188 line.append("/");
3189 line += FRAGMENTPREFIX;
3190 line += TEFACTORSFILE;
3191 output.open(line.c_str(), ios::out);
3192 *out << Verbose(1) << "Saving TEFactors of the total graph ... ";
3193 if(output != NULL) {
3194 for(Graph::iterator runner = KeySetList.begin(); runner != KeySetList.end(); runner++)
3195 output << (*runner).second.second << endl;
3196 *out << Verbose(1) << "done." << endl;
3197 } else {
3198 *out << Verbose(1) << "failed to open " << line << "." << endl;
3199 status = false;
3200 }
3201 output.close();
3202
3203 return status;
3204};
3205
3206/** Storing the bond structure of a molecule to file.
3207 * Simply stores Atom::nr and then the Atom::nr of all bond partners per line.
3208 * \param *out output stream for debugging
3209 * \param *path path to file
3210 * \return true - file written successfully, false - writing failed
3211 */
3212bool molecule::StoreAdjacencyToFile(ofstream *out, char *path)
3213{
3214 ofstream AdjacencyFile;
3215 atom *Walker = NULL;
3216 stringstream line;
3217 bool status = true;
3218
3219 line << path << "/" << FRAGMENTPREFIX << ADJACENCYFILE;
3220 AdjacencyFile.open(line.str().c_str(), ios::out);
3221 *out << Verbose(1) << "Saving adjacency list ... ";
3222 if (AdjacencyFile != NULL) {
3223 Walker = start;
3224 while(Walker->next != end) {
3225 Walker = Walker->next;
3226 AdjacencyFile << Walker->nr << "\t";
3227 for (int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++)
3228 AdjacencyFile << ListOfBondsPerAtom[Walker->nr][i]->GetOtherAtom(Walker)->nr << "\t";
3229 AdjacencyFile << endl;
3230 }
3231 AdjacencyFile.close();
3232 *out << Verbose(1) << "done." << endl;
3233 } else {
3234 *out << Verbose(1) << "failed to open file " << line.str() << "." << endl;
3235 status = false;
3236 }
3237
3238 return status;
3239};
3240
3241/** Checks contents of adjacency file against bond structure in structure molecule.
3242 * \param *out output stream for debugging
3243 * \param *path path to file
3244 * \param **ListOfAtoms allocated (molecule::AtomCount) and filled lookup table for ids (Atom::nr) to *Atom
3245 * \return true - structure is equal, false - not equivalence
3246 */
3247bool molecule::CheckAdjacencyFileAgainstMolecule(ofstream *out, char *path, atom **ListOfAtoms)
3248{
3249 ifstream File;
3250 stringstream filename;
3251 bool status = true;
3252 char *buffer = (char *) Malloc(sizeof(char)*MAXSTRINGSIZE, "molecule::CheckAdjacencyFileAgainstMolecule: *buffer");
3253
3254 filename << path << "/" << FRAGMENTPREFIX << ADJACENCYFILE;
3255 File.open(filename.str().c_str(), ios::out);
3256 *out << Verbose(1) << "Looking at bond structure stored in adjacency file and comparing to present one ... ";
3257 if (File != NULL) {
3258 // allocate storage structure
3259 int NonMatchNumber = 0; // will number of atoms with differing bond structure
3260 int *CurrentBonds = (int *) Malloc(sizeof(int)*8, "molecule::CheckAdjacencyFileAgainstMolecule - CurrentBonds"); // contains parsed bonds of current atom
3261 int CurrentBondsOfAtom;
3262
3263 // Parse the file line by line and count the bonds
3264 while (!File.eof()) {
3265 File.getline(buffer, MAXSTRINGSIZE);
3266 stringstream line;
3267 line.str(buffer);
3268 int AtomNr = -1;
3269 line >> AtomNr;
3270 CurrentBondsOfAtom = -1; // we count one too far due to line end
3271 // parse into structure
3272 if ((AtomNr >= 0) && (AtomNr < AtomCount)) {
3273 while (!line.eof())
3274 line >> CurrentBonds[ ++CurrentBondsOfAtom ];
3275 // compare against present bonds
3276 //cout << Verbose(2) << "Walker is " << *Walker << ", bond partners: ";
3277 if (CurrentBondsOfAtom == NumberOfBondsPerAtom[AtomNr]) {
3278 for(int i=0;i<NumberOfBondsPerAtom[AtomNr];i++) {
3279 int id = ListOfBondsPerAtom[AtomNr][i]->GetOtherAtom(ListOfAtoms[AtomNr])->nr;
3280 int j = 0;
3281 for (;(j<CurrentBondsOfAtom) && (CurrentBonds[j++] != id);); // check against all parsed bonds
3282 if (CurrentBonds[j-1] != id) { // no match ? Then mark in ListOfAtoms
3283 ListOfAtoms[AtomNr] = NULL;
3284 NonMatchNumber++;
3285 status = false;
3286 //out << "[" << id << "]\t";
3287 } else {
3288 //out << id << "\t";
3289 }
3290 }
3291 //out << endl;
3292 } else {
3293 *out << "Number of bonds for Atom " << *ListOfAtoms[AtomNr] << " does not match, parsed " << CurrentBondsOfAtom << " against " << NumberOfBondsPerAtom[AtomNr] << "." << endl;
3294 status = false;
3295 }
3296 }
3297 }
3298 File.close();
3299 File.clear();
3300 if (status) { // if equal we parse the KeySetFile
3301 *out << Verbose(1) << "done: Equal." << endl;
3302 status = true;
3303 } else
3304 *out << Verbose(1) << "done: Not equal by " << NonMatchNumber << " atoms." << endl;
3305 Free((void **)&CurrentBonds, "molecule::CheckAdjacencyFileAgainstMolecule - **CurrentBonds");
3306 } else {
3307 *out << Verbose(1) << "Adjacency file not found." << endl;
3308 status = false;
3309 }
3310 *out << endl;
3311 Free((void **)&buffer, "molecule::CheckAdjacencyFileAgainstMolecule: *buffer");
3312
3313 return status;
3314};
3315
3316/** Checks whether the OrderAtSite is still below \a Order at some site.
3317 * \param *out output stream for debugging
3318 * \param *AtomMask defines true/false per global Atom::nr to mask in/out each nuclear site, used to activate given number of site to increment order adaptively
3319 * \param *GlobalKeySetList list of keysets with global ids (valid in "this" molecule) needed for adaptive increase
3320 * \param Order desired Order if positive, desired exponent in threshold criteria if negative (0 is single-step)
3321 * \param *MinimumRingSize array of max. possible order to avoid loops
3322 * \param *path path to ENERGYPERFRAGMENT file (may be NULL if Order is non-negative)
3323 * \return true - needs further fragmentation, false - does not need fragmentation
3324 */
3325bool molecule::CheckOrderAtSite(ofstream *out, bool *AtomMask, Graph *GlobalKeySetList, int Order, int *MinimumRingSize, char *path)
3326{
3327 atom *Walker = start;
3328 bool status = false;
3329 ifstream InputFile;
3330
3331 // initialize mask list
3332 for(int i=AtomCount;i--;)
3333 AtomMask[i] = false;
3334
3335 if (Order < 0) { // adaptive increase of BondOrder per site
3336 if (AtomMask[AtomCount] == true) // break after one step
3337 return false;
3338 // parse the EnergyPerFragment file
3339 char *buffer = (char *) Malloc(sizeof(char)*MAXSTRINGSIZE, "molecule::CheckOrderAtSite: *buffer");
3340 sprintf(buffer, "%s/%s%s.dat", path, FRAGMENTPREFIX, ENERGYPERFRAGMENT);
3341 InputFile.open(buffer, ios::in);
3342 if ((InputFile != NULL) && (GlobalKeySetList != NULL)) {
3343 // transmorph graph keyset list into indexed KeySetList
3344 map<int,KeySet> IndexKeySetList;
3345 for(Graph::iterator runner = GlobalKeySetList->begin(); runner != GlobalKeySetList->end(); runner++) {
3346 IndexKeySetList.insert( pair<int,KeySet>(runner->second.first,runner->first) );
3347 }
3348 int lines = 0;
3349 // count the number of lines, i.e. the number of fragments
3350 InputFile.getline(buffer, MAXSTRINGSIZE); // skip comment lines
3351 InputFile.getline(buffer, MAXSTRINGSIZE);
3352 while(!InputFile.eof()) {
3353 InputFile.getline(buffer, MAXSTRINGSIZE);
3354 lines++;
3355 }
3356 //*out << Verbose(2) << "Scanned " << lines-1 << " lines." << endl; // one endline too much
3357 InputFile.clear();
3358 InputFile.seekg(ios::beg);
3359 map<int, pair<double,int> > AdaptiveCriteriaList; // (Root No., (Value, Order)) !
3360 int No, FragOrder;
3361 double Value;
3362 // each line represents a fragment root (Atom::nr) id and its energy contribution
3363 InputFile.getline(buffer, MAXSTRINGSIZE); // skip comment lines
3364 InputFile.getline(buffer, MAXSTRINGSIZE);
3365 while(!InputFile.eof()) {
3366 InputFile.getline(buffer, MAXSTRINGSIZE);
3367 if (strlen(buffer) > 2) {
3368 //*out << Verbose(2) << "Scanning: " << buffer << endl;
3369 stringstream line(buffer);
3370 line >> FragOrder;
3371 line >> ws >> No;
3372 line >> ws >> Value; // skip time entry
3373 line >> ws >> Value;
3374 No -= 1; // indices start at 1 in file, not 0
3375 //*out << Verbose(2) << " - yields (" << No << "," << Value << ", " << FragOrder << ")" << endl;
3376
3377 // clean the list of those entries that have been superceded by higher order terms already
3378 map<int,KeySet>::iterator marker = IndexKeySetList.find(No); // find keyset to Frag No.
3379 if (marker != IndexKeySetList.end()) { // if found
3380 Value *= 1 + MYEPSILON*(*((*marker).second.begin())); // in case of equal energies this makes em not equal without changing anything actually
3381 // as the smallest number in each set has always been the root (we use global id to keep the doubles away), seek smallest and insert into AtomMask
3382 pair <map<int, pair<double,int> >::iterator, bool> InsertedElement = AdaptiveCriteriaList.insert( make_pair(*((*marker).second.begin()), pair<double,int>( fabs(Value), FragOrder) ));
3383 map<int, pair<double,int> >::iterator PresentItem = InsertedElement.first;
3384 if (!InsertedElement.second) { // this root is already present
3385 if ((*PresentItem).second.second < FragOrder) // if order there is lower, update entry with higher-order term
3386 //if ((*PresentItem).second.first < (*runner).first) // as higher-order terms are not always better, we skip this part (which would always include this site into adaptive increase)
3387 { // if value is smaller, update value and order
3388 (*PresentItem).second.first = fabs(Value);
3389 (*PresentItem).second.second = FragOrder;
3390 *out << Verbose(2) << "Updated element (" << (*PresentItem).first << ",[" << (*PresentItem).second.first << "," << (*PresentItem).second.second << "])." << endl;
3391 } else {
3392 *out << Verbose(2) << "Did not update element " << (*PresentItem).first << " as " << FragOrder << " is less than or equal to " << (*PresentItem).second.second << "." << endl;
3393 }
3394 } else {
3395 *out << Verbose(2) << "Inserted element (" << (*PresentItem).first << ",[" << (*PresentItem).second.first << "," << (*PresentItem).second.second << "])." << endl;
3396 }
3397 } else {
3398 *out << Verbose(1) << "No Fragment under No. " << No << "found." << endl;
3399 }
3400 }
3401 }
3402 // then map back onto (Value, (Root Nr., Order)) (i.e. sorted by value to pick the highest ones)
3403 map<double, pair<int,int> > FinalRootCandidates;
3404 *out << Verbose(1) << "Root candidate list is: " << endl;
3405 for(map<int, pair<double,int> >::iterator runner = AdaptiveCriteriaList.begin(); runner != AdaptiveCriteriaList.end(); runner++) {
3406 Walker = FindAtom((*runner).first);
3407 if (Walker != NULL) {
3408 //if ((*runner).second.second >= Walker->AdaptiveOrder) { // only insert if this is an "active" root site for the current order
3409 if (!Walker->MaxOrder) {
3410 *out << Verbose(2) << "(" << (*runner).first << ",[" << (*runner).second.first << "," << (*runner).second.second << "])" << endl;
3411 FinalRootCandidates.insert( make_pair( (*runner).second.first, pair<int,int>((*runner).first, (*runner).second.second) ) );
3412 } else {
3413 *out << Verbose(2) << "Excluding (" << *Walker << ", " << (*runner).first << ",[" << (*runner).second.first << "," << (*runner).second.second << "]), as it has reached its maximum order." << endl;
3414 }
3415 } else {
3416 cerr << "Atom No. " << (*runner).second.first << " was not found in this molecule." << endl;
3417 }
3418 }
3419 // pick the ones still below threshold and mark as to be adaptively updated
3420 for(map<double, pair<int,int> >::iterator runner = FinalRootCandidates.upper_bound(pow(10.,Order)); runner != FinalRootCandidates.end(); runner++) {
3421 No = (*runner).second.first;
3422 Walker = FindAtom(No);
3423 //if (Walker->AdaptiveOrder < MinimumRingSize[Walker->nr]) {
3424 *out << Verbose(2) << "Root " << No << " is still above threshold (10^{" << Order <<"}: " << runner->first << ", setting entry " << No << " of Atom mask to true." << endl;
3425 AtomMask[No] = true;
3426 status = true;
3427 //} else
3428 //*out << Verbose(2) << "Root " << No << " is still above threshold (10^{" << Order <<"}: " << runner->first << ", however MinimumRingSize of " << MinimumRingSize[Walker->nr] << " does not allow further adaptive increase." << endl;
3429 }
3430 // close and done
3431 InputFile.close();
3432 InputFile.clear();
3433 } else {
3434 cerr << "Unable to parse " << buffer << " file, incrementing all." << endl;
3435 while (Walker->next != end) {
3436 Walker = Walker->next;
3437 #ifdef ADDHYDROGEN
3438 if (Walker->type->Z != 1) // skip hydrogen
3439 #endif
3440 {
3441 AtomMask[Walker->nr] = true; // include all (non-hydrogen) atoms
3442 status = true;
3443 }
3444 }
3445 }
3446 Free((void **)&buffer, "molecule::CheckOrderAtSite: *buffer");
3447 // pick a given number of highest values and set AtomMask
3448 } else { // global increase of Bond Order
3449 while (Walker->next != end) {
3450 Walker = Walker->next;
3451 #ifdef ADDHYDROGEN
3452 if (Walker->type->Z != 1) // skip hydrogen
3453 #endif
3454 {
3455 AtomMask[Walker->nr] = true; // include all (non-hydrogen) atoms
3456 if ((Order != 0) && (Walker->AdaptiveOrder < Order)) // && (Walker->AdaptiveOrder < MinimumRingSize[Walker->nr]))
3457 status = true;
3458 }
3459 }
3460 if ((Order == 0) && (AtomMask[AtomCount] == false)) // single stepping, just check
3461 status = true;
3462
3463 if (!status) {
3464 if (Order == 0)
3465 *out << Verbose(1) << "Single stepping done." << endl;
3466 else
3467 *out << Verbose(1) << "Order at every site is already equal or above desired order " << Order << "." << endl;
3468 }
3469 }
3470
3471 // print atom mask for debugging
3472 *out << " ";
3473 for(int i=0;i<AtomCount;i++)
3474 *out << (i % 10);
3475 *out << endl << "Atom mask is: ";
3476 for(int i=0;i<AtomCount;i++)
3477 *out << (AtomMask[i] ? "t" : "f");
3478 *out << endl;
3479
3480 return status;
3481};
3482
3483/** Create a SortIndex to map from atomic labels to the sequence in which the atoms are given in the config file.
3484 * \param *out output stream for debugging
3485 * \param *&SortIndex Mapping array of size molecule::AtomCount
3486 * \return true - success, false - failure of SortIndex alloc
3487 */
3488bool molecule::CreateMappingLabelsToConfigSequence(ofstream *out, int *&SortIndex)
3489{
3490 element *runner = elemente->start;
3491 int AtomNo = 0;
3492 atom *Walker = NULL;
3493
3494 if (SortIndex != NULL) {
3495 *out << Verbose(1) << "SortIndex is " << SortIndex << " and not NULL as expected." << endl;
3496 return false;
3497 }
3498 SortIndex = (int *) Malloc(sizeof(int)*AtomCount, "molecule::FragmentMolecule: *SortIndex");
3499 for(int i=AtomCount;i--;)
3500 SortIndex[i] = -1;
3501 while (runner->next != elemente->end) { // go through every element
3502 runner = runner->next;
3503 if (ElementsInMolecule[runner->Z]) { // if this element got atoms
3504 Walker = start;
3505 while (Walker->next != end) { // go through every atom of this element
3506 Walker = Walker->next;
3507 if (Walker->type->Z == runner->Z) // if this atom fits to element
3508 SortIndex[Walker->nr] = AtomNo++;
3509 }
3510 }
3511 }
3512 return true;
3513};
3514
3515/** Performs a many-body bond order analysis for a given bond order.
3516 * -# parses adjacency, keysets and orderatsite files
3517 * -# performs DFS to find connected subgraphs (to leave this in was a design decision: might be useful later)
3518 * -# RootStack is created for every subgraph (here, later we implement the "update 10 sites with highest energ
3519y contribution", and that's why this consciously not done in the following loop)
3520 * -# in a loop over all subgraphs
3521 * -# calls FragmentBOSSANOVA with this RootStack and within the subgraph molecule structure
3522 * -# creates molecule (fragment)s from the returned keysets (StoreFragmentFromKeySet)
3523 * -# combines the generated molecule lists from all subgraphs
3524 * -# saves to disk: fragment configs, adjacency, orderatsite, keyset files
3525 * Note that as we split "this" molecule up into a list of subgraphs, i.e. a MoleculeListClass, we have two sets
3526 * of vertex indices: Global always means the index in "this" molecule, whereas local refers to the molecule or
3527 * subgraph in the MoleculeListClass.
3528 * \param *out output stream for debugging
3529 * \param Order up to how many neighbouring bonds a fragment contains in BondOrderScheme::BottumUp scheme
3530 * \param *configuration configuration for writing config files for each fragment
3531 * \return 1 - continue, 2 - stop (no fragmentation occured)
3532 */
3533int molecule::FragmentMolecule(ofstream *out, int Order, config *configuration)
3534{
3535 MoleculeListClass *BondFragments = NULL;
3536 int *SortIndex = NULL;
3537 int *MinimumRingSize = NULL;
3538 int FragmentCounter;
3539 MoleculeLeafClass *MolecularWalker = NULL;
3540 MoleculeLeafClass *Subgraphs = NULL; // list of subgraphs from DFS analysis
3541 fstream File;
3542 bool FragmentationToDo = true;
3543 bool CheckOrder = false;
3544 Graph **FragmentList = NULL;
3545 Graph *ParsedFragmentList = NULL;
3546 Graph TotalGraph; // graph with all keysets however local numbers
3547 int TotalNumberOfKeySets = 0;
3548 atom **ListOfAtoms = NULL;
3549 atom ***ListOfLocalAtoms = NULL;
3550 bool *AtomMask = NULL;
3551
3552 *out << endl;
3553#ifdef ADDHYDROGEN
3554 *out << Verbose(0) << "I will treat hydrogen special and saturate dangling bonds with it." << endl;
3555#else
3556 *out << Verbose(0) << "Hydrogen is treated just like the rest of the lot." << endl;
3557#endif
3558
3559 // ++++++++++++++++++++++++++++ INITIAL STUFF: Bond structure analysis, file parsing, ... ++++++++++++++++++++++++++++++++++++++++++
3560
3561 // ===== 1. Check whether bond structure is same as stored in files ====
3562
3563 // fill the adjacency list
3564 CreateListOfBondsPerAtom(out);
3565
3566 // create lookup table for Atom::nr
3567 FragmentationToDo = FragmentationToDo && CreateFatherLookupTable(out, start, end, ListOfAtoms, AtomCount);
3568
3569 // === compare it with adjacency file ===
3570 FragmentationToDo = FragmentationToDo && CheckAdjacencyFileAgainstMolecule(out, configuration->configpath, ListOfAtoms);
3571 Free((void **)&ListOfAtoms, "molecule::FragmentMolecule - **ListOfAtoms");
3572
3573 // ===== 2. perform a DFS analysis to gather info on cyclic structure and a list of disconnected subgraphs =====
3574 Subgraphs = DepthFirstSearchAnalysis(out, MinimumRingSize);
3575 // fill the bond structure of the individually stored subgraphs
3576 Subgraphs->next->FillBondStructureFromReference(out, this, (FragmentCounter = 0), ListOfLocalAtoms, false); // we want to keep the created ListOfLocalAtoms
3577
3578 // ===== 3. if structure still valid, parse key set file and others =====
3579 FragmentationToDo = FragmentationToDo && ParseKeySetFile(out, configuration->configpath, ParsedFragmentList);
3580
3581 // ===== 4. check globally whether there's something to do actually (first adaptivity check)
3582 FragmentationToDo = FragmentationToDo && ParseOrderAtSiteFromFile(out, configuration->configpath);
3583
3584 // =================================== Begin of FRAGMENTATION ===============================
3585 // ===== 6a. assign each keyset to its respective subgraph =====
3586 Subgraphs->next->AssignKeySetsToFragment(out, this, ParsedFragmentList, ListOfLocalAtoms, FragmentList, (FragmentCounter = 0), false);
3587
3588 // ===== 6b. prepare and go into the adaptive (Order<0), single-step (Order==0) or incremental (Order>0) cycle
3589 KeyStack *RootStack = new KeyStack[Subgraphs->next->Count()];
3590 AtomMask = new bool[AtomCount+1];
3591 AtomMask[AtomCount] = false;
3592 FragmentationToDo = false; // if CheckOrderAtSite just ones recommends fragmentation, we will save fragments afterwards
3593 while ((CheckOrder = CheckOrderAtSite(out, AtomMask, ParsedFragmentList, Order, MinimumRingSize, configuration->configpath))) {
3594 FragmentationToDo = FragmentationToDo || CheckOrder;
3595 AtomMask[AtomCount] = true; // last plus one entry is used as marker that we have been through this loop once already in CheckOrderAtSite()
3596 // ===== 6b. fill RootStack for each subgraph (second adaptivity check) =====
3597 Subgraphs->next->FillRootStackForSubgraphs(out, RootStack, AtomMask, (FragmentCounter = 0));
3598
3599 // ===== 7. fill the bond fragment list =====
3600 FragmentCounter = 0;
3601 MolecularWalker = Subgraphs;
3602 while (MolecularWalker->next != NULL) {
3603 MolecularWalker = MolecularWalker->next;
3604 *out << Verbose(1) << "Fragmenting subgraph " << MolecularWalker << "." << endl;
3605 // output ListOfBondsPerAtom for debugging
3606 MolecularWalker->Leaf->OutputListOfBonds(out);
3607 if (MolecularWalker->Leaf->first->next != MolecularWalker->Leaf->last) {
3608
3609 // call BOSSANOVA method
3610 *out << Verbose(0) << endl << " ========== BOND ENERGY of subgraph " << FragmentCounter << " ========================= " << endl;
3611 MolecularWalker->Leaf->FragmentBOSSANOVA(out, FragmentList[FragmentCounter], RootStack[FragmentCounter], MinimumRingSize);
3612 } else {
3613 cerr << "Subgraph " << MolecularWalker << " has no atoms!" << endl;
3614 }
3615 FragmentCounter++; // next fragment list
3616 }
3617 }
3618 delete[](RootStack);
3619 delete[](AtomMask);
3620 delete(ParsedFragmentList);
3621 delete[](MinimumRingSize);
3622
3623 // free the index lookup list
3624 for (int i=FragmentCounter;i--;)
3625 Free((void **)&ListOfLocalAtoms[i], "molecule::FragmentMolecule - *ListOfLocalAtoms[]");
3626 Free((void **)&ListOfLocalAtoms, "molecule::FragmentMolecule - **ListOfLocalAtoms");
3627
3628 // ==================================== End of FRAGMENTATION ============================================
3629
3630 // ===== 8a. translate list into global numbers (i.e. ones that are valid in "this" molecule, not in MolecularWalker->Leaf)
3631 Subgraphs->next->TranslateIndicesToGlobalIDs(out, FragmentList, (FragmentCounter = 0), TotalNumberOfKeySets, TotalGraph);
3632
3633 // free subgraph memory again
3634 FragmentCounter = 0;
3635 if (Subgraphs != NULL) {
3636 while (Subgraphs->next != NULL) {
3637 Subgraphs = Subgraphs->next;
3638 delete(FragmentList[FragmentCounter++]);
3639 delete(Subgraphs->previous);
3640 }
3641 delete(Subgraphs);
3642 }
3643 Free((void **)&FragmentList, "molecule::FragmentMolecule - **FragmentList");
3644
3645 // ===== 8b. gather keyset lists (graphs) from all subgraphs and transform into MoleculeListClass =====
3646 //if (FragmentationToDo) { // we should always store the fragments again as coordination might have changed slightly without changing bond structure
3647 // allocate memory for the pointer array and transmorph graphs into full molecular fragments
3648 BondFragments = new MoleculeListClass(TotalGraph.size(), AtomCount);
3649 int k=0;
3650 for(Graph::iterator runner = TotalGraph.begin(); runner != TotalGraph.end(); runner++) {
3651 KeySet test = (*runner).first;
3652 *out << "Fragment No." << (*runner).second.first << " with TEFactor " << (*runner).second.second << "." << endl;
3653 BondFragments->ListOfMolecules[k] = StoreFragmentFromKeySet(out, test, configuration);
3654 k++;
3655 }
3656 *out << k << "/" << BondFragments->NumberOfMolecules << " fragments generated from the keysets." << endl;
3657
3658 // ===== 9. Save fragments' configuration and keyset files et al to disk ===
3659 if (BondFragments->NumberOfMolecules != 0) {
3660 // create the SortIndex from BFS labels to order in the config file
3661 CreateMappingLabelsToConfigSequence(out, SortIndex);
3662
3663 *out << Verbose(1) << "Writing " << BondFragments->NumberOfMolecules << " possible bond fragmentation configs" << endl;
3664 if (BondFragments->OutputConfigForListOfFragments(out, FRAGMENTPREFIX, configuration, SortIndex, true, true))
3665 *out << Verbose(1) << "All configs written." << endl;
3666 else
3667 *out << Verbose(1) << "Some config writing failed." << endl;
3668
3669 // store force index reference file
3670 BondFragments->StoreForcesFile(out, configuration->configpath, SortIndex);
3671
3672 // store keysets file
3673 StoreKeySetFile(out, TotalGraph, configuration->configpath);
3674
3675 // store Adjacency file
3676 StoreAdjacencyToFile(out, configuration->configpath);
3677
3678 // store Hydrogen saturation correction file
3679 BondFragments->AddHydrogenCorrection(out, configuration->configpath);
3680
3681 // store adaptive orders into file
3682 StoreOrderAtSiteFile(out, configuration->configpath);
3683
3684 // restore orbital and Stop values
3685 CalculateOrbitals(*configuration);
3686
3687 // free memory for bond part
3688 *out << Verbose(1) << "Freeing bond memory" << endl;
3689 delete(FragmentList); // remove bond molecule from memory
3690 Free((void **)&SortIndex, "molecule::FragmentMolecule: *SortIndex");
3691 } else
3692 *out << Verbose(1) << "FragmentList is zero on return, splitting failed." << endl;
3693 //} else
3694 // *out << Verbose(1) << "No fragments to store." << endl;
3695 *out << Verbose(0) << "End of bond fragmentation." << endl;
3696
3697 return ((int)(!FragmentationToDo)+1); // 1 - continue, 2 - stop (no fragmentation occured)
3698};
3699
3700/** Stores pairs (Atom::nr, Atom::AdaptiveOrder) into file.
3701 * Atoms not present in the file get "-1".
3702 * \param *out output stream for debugging
3703 * \param *path path to file ORDERATSITEFILE
3704 * \return true - file writable, false - not writable
3705 */
3706bool molecule::StoreOrderAtSiteFile(ofstream *out, char *path)
3707{
3708 stringstream line;
3709 ofstream file;
3710
3711 line << path << "/" << FRAGMENTPREFIX << ORDERATSITEFILE;
3712 file.open(line.str().c_str());
3713 *out << Verbose(1) << "Writing OrderAtSite " << ORDERATSITEFILE << " ... " << endl;
3714 if (file != NULL) {
3715 atom *Walker = start;
3716 while (Walker->next != end) {
3717 Walker = Walker->next;
3718 file << Walker->nr << "\t" << (int)Walker->AdaptiveOrder << "\t" << (int)Walker->MaxOrder << endl;
3719 *out << Verbose(2) << "Storing: " << Walker->nr << "\t" << (int)Walker->AdaptiveOrder << "\t" << (int)Walker->MaxOrder << "." << endl;
3720 }
3721 file.close();
3722 *out << Verbose(1) << "done." << endl;
3723 return true;
3724 } else {
3725 *out << Verbose(1) << "failed to open file " << line.str() << "." << endl;
3726 return false;
3727 }
3728};
3729
3730/** Parses pairs(Atom::nr, Atom::AdaptiveOrder) from file and stores in molecule's Atom's.
3731 * Atoms not present in the file get "0".
3732 * \param *out output stream for debugging
3733 * \param *path path to file ORDERATSITEFILEe
3734 * \return true - file found and scanned, false - file not found
3735 * \sa ParseKeySetFile() and CheckAdjacencyFileAgainstMolecule() as this is meant to be used in conjunction with the two
3736 */
3737bool molecule::ParseOrderAtSiteFromFile(ofstream *out, char *path)
3738{
3739 unsigned char *OrderArray = (unsigned char *) Malloc(sizeof(unsigned char)*AtomCount, "molecule::ParseOrderAtSiteFromFile - *OrderArray");
3740 bool *MaxArray = (bool *) Malloc(sizeof(bool)*AtomCount, "molecule::ParseOrderAtSiteFromFile - *MaxArray");
3741 bool status;
3742 int AtomNr, value;
3743 stringstream line;
3744 ifstream file;
3745
3746 *out << Verbose(1) << "Begin of ParseOrderAtSiteFromFile" << endl;
3747 for(int i=AtomCount;i--;)
3748 OrderArray[i] = 0;
3749 line << path << "/" << FRAGMENTPREFIX << ORDERATSITEFILE;
3750 file.open(line.str().c_str());
3751 if (file != NULL) {
3752 for (int i=AtomCount;i--;) { // initialise with 0
3753 OrderArray[i] = 0;
3754 MaxArray[i] = 0;
3755 }
3756 while (!file.eof()) { // parse from file
3757 AtomNr = -1;
3758 file >> AtomNr;
3759 if (AtomNr != -1) { // test whether we really parsed something (this is necessary, otherwise last atom is set twice and to 0 on second time)
3760 file >> value;
3761 OrderArray[AtomNr] = value;
3762 file >> value;
3763 MaxArray[AtomNr] = value;
3764 //*out << Verbose(2) << "AtomNr " << AtomNr << " with order " << (int)OrderArray[AtomNr] << " and max order set to " << (int)MaxArray[AtomNr] << "." << endl;
3765 }
3766 }
3767 atom *Walker = start;
3768 while (Walker->next != end) { // fill into atom classes
3769 Walker = Walker->next;
3770 Walker->AdaptiveOrder = OrderArray[Walker->nr];
3771 Walker->MaxOrder = MaxArray[Walker->nr];
3772 *out << Verbose(2) << *Walker << " gets order " << (int)Walker->AdaptiveOrder << " and is " << (!Walker->MaxOrder ? "not " : " ") << "maxed." << endl;
3773 }
3774 file.close();
3775 *out << Verbose(1) << "done." << endl;
3776 status = true;
3777 } else {
3778 *out << Verbose(1) << "failed to open file " << line.str() << "." << endl;
3779 status = false;
3780 }
3781 Free((void **)&OrderArray, "molecule::ParseOrderAtSiteFromFile - *OrderArray");
3782 Free((void **)&MaxArray, "molecule::ParseOrderAtSiteFromFile - *MaxArray");
3783
3784 *out << Verbose(1) << "End of ParseOrderAtSiteFromFile" << endl;
3785 return status;
3786};
3787
3788/** Creates an 2d array of pointer with an entry for each atom and each bond it has.
3789 * Updates molecule::ListOfBondsPerAtom, molecule::NumberOfBondsPerAtom by parsing through
3790 * bond chain list, using molecule::AtomCount and molecule::BondCount.
3791 * Allocates memory, fills the array and exits
3792 * \param *out output stream for debugging
3793 */
3794void molecule::CreateListOfBondsPerAtom(ofstream *out)
3795{
3796 bond *Binder = NULL;
3797 atom *Walker = NULL;
3798 int TotalDegree;
3799 *out << Verbose(1) << "Begin of Creating ListOfBondsPerAtom: AtomCount = " << AtomCount << "\tBondCount = " << BondCount << "\tNoNonBonds = " << NoNonBonds << "." << endl;
3800
3801 // re-allocate memory
3802 *out << Verbose(2) << "(Re-)Allocating memory." << endl;
3803 if (ListOfBondsPerAtom != NULL) {
3804 for(int i=AtomCount;i--;)
3805 Free((void **)&ListOfBondsPerAtom[i], "molecule::CreateListOfBondsPerAtom: ListOfBondsPerAtom[i]");
3806 Free((void **)&ListOfBondsPerAtom, "molecule::CreateListOfBondsPerAtom: ListOfBondsPerAtom");
3807 }
3808 if (NumberOfBondsPerAtom != NULL)
3809 Free((void **)&NumberOfBondsPerAtom, "molecule::CreateListOfBondsPerAtom: NumberOfBondsPerAtom");
3810 ListOfBondsPerAtom = (bond ***) Malloc(sizeof(bond **)*AtomCount, "molecule::CreateListOfBondsPerAtom: ***ListOfBondsPerAtom");
3811 NumberOfBondsPerAtom = (int *) Malloc(sizeof(int)*AtomCount, "molecule::CreateListOfBondsPerAtom: *NumberOfBondsPerAtom");
3812
3813 // reset bond counts per atom
3814 for(int i=AtomCount;i--;)
3815 NumberOfBondsPerAtom[i] = 0;
3816 // count bonds per atom
3817 Binder = first;
3818 while (Binder->next != last) {
3819 Binder = Binder->next;
3820 NumberOfBondsPerAtom[Binder->leftatom->nr]++;
3821 NumberOfBondsPerAtom[Binder->rightatom->nr]++;
3822 }
3823 for(int i=AtomCount;i--;) {
3824 // allocate list of bonds per atom
3825 ListOfBondsPerAtom[i] = (bond **) Malloc(sizeof(bond *)*NumberOfBondsPerAtom[i], "molecule::CreateListOfBondsPerAtom: **ListOfBondsPerAtom[]");
3826 // clear the list again, now each NumberOfBondsPerAtom marks current free field
3827 NumberOfBondsPerAtom[i] = 0;
3828 }
3829 // fill the list
3830 Binder = first;
3831 while (Binder->next != last) {
3832 Binder = Binder->next;
3833 ListOfBondsPerAtom[Binder->leftatom->nr][NumberOfBondsPerAtom[Binder->leftatom->nr]++] = Binder;
3834 ListOfBondsPerAtom[Binder->rightatom->nr][NumberOfBondsPerAtom[Binder->rightatom->nr]++] = Binder;
3835 }
3836
3837 // output list for debugging
3838 *out << Verbose(3) << "ListOfBondsPerAtom for each atom:" << endl;
3839 Walker = start;
3840 while (Walker->next != end) {
3841 Walker = Walker->next;
3842 *out << Verbose(4) << "Atom " << Walker->Name << " with " << NumberOfBondsPerAtom[Walker->nr] << " bonds: ";
3843 TotalDegree = 0;
3844 for (int j=0;j<NumberOfBondsPerAtom[Walker->nr];j++) {
3845 *out << *ListOfBondsPerAtom[Walker->nr][j] << "\t";
3846 TotalDegree += ListOfBondsPerAtom[Walker->nr][j]->BondDegree;
3847 }
3848 *out << " -- TotalDegree: " << TotalDegree << endl;
3849 }
3850 *out << Verbose(1) << "End of Creating ListOfBondsPerAtom." << endl << endl;
3851};
3852
3853/** Adds atoms up to \a BondCount distance from \a *Root and notes them down in \a **AddedAtomList.
3854 * Gray vertices are always enqueued in an StackClass<atom *> FIFO queue, the rest is usual BFS with adding vertices found was
3855 * white and putting into queue.
3856 * \param *out output stream for debugging
3857 * \param *Mol Molecule class to add atoms to
3858 * \param **AddedAtomList list with added atom pointers, index is atom father's number
3859 * \param **AddedBondList list with added bond pointers, index is bond father's number
3860 * \param *Root root vertex for BFS
3861 * \param *Bond bond not to look beyond
3862 * \param BondOrder maximum distance for vertices to add
3863 * \param IsAngstroem lengths are in angstroem or bohrradii
3864 */
3865void molecule::BreadthFirstSearchAdd(ofstream *out, molecule *Mol, atom **&AddedAtomList, bond **&AddedBondList, atom *Root, bond *Bond, int BondOrder, bool IsAngstroem)
3866{
3867 atom **PredecessorList = (atom **) Malloc(sizeof(atom *)*AtomCount, "molecule::BreadthFirstSearchAdd: **PredecessorList");
3868 int *ShortestPathList = (int *) Malloc(sizeof(int)*AtomCount, "molecule::BreadthFirstSearchAdd: *ShortestPathList");
3869 enum Shading *ColorList = (enum Shading *) Malloc(sizeof(enum Shading)*AtomCount, "molecule::BreadthFirstSearchAdd: *ColorList");
3870 class StackClass<atom *> *AtomStack = new StackClass<atom *>(AtomCount);
3871 atom *Walker = NULL, *OtherAtom = NULL;
3872 bond *Binder = NULL;
3873
3874 // add Root if not done yet
3875 AtomStack->ClearStack();
3876 if (AddedAtomList[Root->nr] == NULL) // add Root if not yet present
3877 AddedAtomList[Root->nr] = Mol->AddCopyAtom(Root);
3878 AtomStack->Push(Root);
3879
3880 // initialise each vertex as white with no predecessor, empty queue, color Root lightgray
3881 for (int i=AtomCount;i--;) {
3882 PredecessorList[i] = NULL;
3883 ShortestPathList[i] = -1;
3884 if (AddedAtomList[i] != NULL) // mark already present atoms (i.e. Root and maybe others) as visited
3885 ColorList[i] = lightgray;
3886 else
3887 ColorList[i] = white;
3888 }
3889 ShortestPathList[Root->nr] = 0;
3890
3891 // and go on ... Queue always contains all lightgray vertices
3892 while (!AtomStack->IsEmpty()) {
3893 // we have to pop the oldest atom from stack. This keeps the atoms on the stack always of the same ShortestPath distance.
3894 // e.g. if current atom is 2, push to end of stack are of length 3, but first all of length 2 would be popped. They again
3895 // append length of 3 (their neighbours). Thus on stack we have always atoms of a certain length n at bottom of stack and
3896 // followed by n+1 till top of stack.
3897 Walker = AtomStack->PopFirst(); // pop oldest added
3898 *out << Verbose(1) << "Current Walker is: " << Walker->Name << ", and has " << NumberOfBondsPerAtom[Walker->nr] << " bonds." << endl;
3899 for(int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++) {
3900 Binder = ListOfBondsPerAtom[Walker->nr][i];
3901 if (Binder != NULL) { // don't look at bond equal NULL
3902 OtherAtom = Binder->GetOtherAtom(Walker);
3903 *out << Verbose(2) << "Current OtherAtom is: " << OtherAtom->Name << " for bond " << *Binder << "." << endl;
3904 if (ColorList[OtherAtom->nr] == white) {
3905 if (Binder != Bond) // let other atom white if it's via Root bond. In case it's cyclic it has to be reached again (yet Root is from OtherAtom already black, thus no problem)
3906 ColorList[OtherAtom->nr] = lightgray;
3907 PredecessorList[OtherAtom->nr] = Walker; // Walker is the predecessor
3908 ShortestPathList[OtherAtom->nr] = ShortestPathList[Walker->nr]+1;
3909 *out << Verbose(2) << "Coloring OtherAtom " << OtherAtom->Name << " " << ((ColorList[OtherAtom->nr] == white) ? "white" : "lightgray") << ", its predecessor is " << Walker->Name << " and its Shortest Path is " << ShortestPathList[OtherAtom->nr] << " egde(s) long." << endl;
3910 if ((((ShortestPathList[OtherAtom->nr] < BondOrder) && (Binder != Bond))) ) { // Check for maximum distance
3911 *out << Verbose(3);
3912 if (AddedAtomList[OtherAtom->nr] == NULL) { // add if it's not been so far
3913 AddedAtomList[OtherAtom->nr] = Mol->AddCopyAtom(OtherAtom);
3914 *out << "Added OtherAtom " << OtherAtom->Name;
3915 AddedBondList[Binder->nr] = Mol->AddBond(AddedAtomList[Walker->nr], AddedAtomList[OtherAtom->nr], Binder->BondDegree);
3916 AddedBondList[Binder->nr]->Cyclic = Binder->Cyclic;
3917 AddedBondList[Binder->nr]->Type = Binder->Type;
3918 *out << " and bond " << *(AddedBondList[Binder->nr]) << ", ";
3919 } else { // this code should actually never come into play (all white atoms are not yet present in BondMolecule, that's why they are white in the first place)
3920 *out << "Not adding OtherAtom " << OtherAtom->Name;
3921 if (AddedBondList[Binder->nr] == NULL) {
3922 AddedBondList[Binder->nr] = Mol->AddBond(AddedAtomList[Walker->nr], AddedAtomList[OtherAtom->nr], Binder->BondDegree);
3923 AddedBondList[Binder->nr]->Cyclic = Binder->Cyclic;
3924 AddedBondList[Binder->nr]->Type = Binder->Type;
3925 *out << ", added Bond " << *(AddedBondList[Binder->nr]);
3926 } else
3927 *out << ", not added Bond ";
3928 }
3929 *out << ", putting OtherAtom into queue." << endl;
3930 AtomStack->Push(OtherAtom);
3931 } else { // out of bond order, then replace
3932 if ((AddedAtomList[OtherAtom->nr] == NULL) && (Binder->Cyclic))
3933 ColorList[OtherAtom->nr] = white; // unmark if it has not been queued/added, to make it available via its other bonds (cyclic)
3934 if (Binder == Bond)
3935 *out << Verbose(3) << "Not Queueing, is the Root bond";
3936 else if (ShortestPathList[OtherAtom->nr] >= BondOrder)
3937 *out << Verbose(3) << "Not Queueing, is out of Bond Count of " << BondOrder;
3938 if (!Binder->Cyclic)
3939 *out << ", is not part of a cyclic bond, saturating bond with Hydrogen." << endl;
3940 if (AddedBondList[Binder->nr] == NULL) {
3941 if ((AddedAtomList[OtherAtom->nr] != NULL)) { // .. whether we add or saturate
3942 AddedBondList[Binder->nr] = Mol->AddBond(AddedAtomList[Walker->nr], AddedAtomList[OtherAtom->nr], Binder->BondDegree);
3943 AddedBondList[Binder->nr]->Cyclic = Binder->Cyclic;
3944 AddedBondList[Binder->nr]->Type = Binder->Type;
3945 } else {
3946#ifdef ADDHYDROGEN
3947 Mol->AddHydrogenReplacementAtom(out, Binder, AddedAtomList[Walker->nr], Walker, OtherAtom, ListOfBondsPerAtom[Walker->nr], NumberOfBondsPerAtom[Walker->nr], IsAngstroem);
3948#endif
3949 }
3950 }
3951 }
3952 } else {
3953 *out << Verbose(3) << "Not Adding, has already been visited." << endl;
3954 // This has to be a cyclic bond, check whether it's present ...
3955 if (AddedBondList[Binder->nr] == NULL) {
3956 if ((Binder != Bond) && (Binder->Cyclic) && (((ShortestPathList[Walker->nr]+1) < BondOrder))) {
3957 AddedBondList[Binder->nr] = Mol->AddBond(AddedAtomList[Walker->nr], AddedAtomList[OtherAtom->nr], Binder->BondDegree);
3958 AddedBondList[Binder->nr]->Cyclic = Binder->Cyclic;
3959 AddedBondList[Binder->nr]->Type = Binder->Type;
3960 } else { // if it's root bond it has to broken (otherwise we would not create the fragments)
3961#ifdef ADDHYDROGEN
3962 Mol->AddHydrogenReplacementAtom(out, Binder, AddedAtomList[Walker->nr], Walker, OtherAtom, ListOfBondsPerAtom[Walker->nr], NumberOfBondsPerAtom[Walker->nr], IsAngstroem);
3963#endif
3964 }
3965 }
3966 }
3967 }
3968 }
3969 ColorList[Walker->nr] = black;
3970 *out << Verbose(1) << "Coloring Walker " << Walker->Name << " black." << endl;
3971 }
3972 Free((void **)&PredecessorList, "molecule::BreadthFirstSearchAdd: **PredecessorList");
3973 Free((void **)&ShortestPathList, "molecule::BreadthFirstSearchAdd: **ShortestPathList");
3974 Free((void **)&ColorList, "molecule::BreadthFirstSearchAdd: **ColorList");
3975 delete(AtomStack);
3976};
3977
3978/** Adds bond structure to this molecule from \a Father molecule.
3979 * This basically causes this molecule to become an induced subgraph of the \a Father, i.e. for every bond in Father
3980 * with end points present in this molecule, bond is created in this molecule.
3981 * Special care was taken to ensure that this is of complexity O(N), where N is the \a Father's molecule::AtomCount.
3982 * \param *out output stream for debugging
3983 * \param *Father father molecule
3984 * \return true - is induced subgraph, false - there are atoms with fathers not in \a Father
3985 * \todo not checked, not fully working probably
3986 */
3987bool molecule::BuildInducedSubgraph(ofstream *out, const molecule *Father)
3988{
3989 atom *Walker = NULL, *OtherAtom = NULL;
3990 bool status = true;
3991 atom **ParentList = (atom **) Malloc(sizeof(atom *)*Father->AtomCount, "molecule::BuildInducedSubgraph: **ParentList");
3992
3993 *out << Verbose(2) << "Begin of BuildInducedSubgraph." << endl;
3994
3995 // reset parent list
3996 *out << Verbose(3) << "Resetting ParentList." << endl;
3997 for (int i=Father->AtomCount;i--;)
3998 ParentList[i] = NULL;
3999
4000 // fill parent list with sons
4001 *out << Verbose(3) << "Filling Parent List." << endl;
4002 Walker = start;
4003 while (Walker->next != end) {
4004 Walker = Walker->next;
4005 ParentList[Walker->father->nr] = Walker;
4006 // Outputting List for debugging
4007 *out << Verbose(4) << "Son["<< Walker->father->nr <<"] of " << Walker->father << " is " << ParentList[Walker->father->nr] << "." << endl;
4008 }
4009
4010 // check each entry of parent list and if ok (one-to-and-onto matching) create bonds
4011 *out << Verbose(3) << "Creating bonds." << endl;
4012 Walker = Father->start;
4013 while (Walker->next != Father->end) {
4014 Walker = Walker->next;
4015 if (ParentList[Walker->nr] != NULL) {
4016 if (ParentList[Walker->nr]->father != Walker) {
4017 status = false;
4018 } else {
4019 for (int i=0;i<Father->NumberOfBondsPerAtom[Walker->nr];i++) {
4020 OtherAtom = Father->ListOfBondsPerAtom[Walker->nr][i]->GetOtherAtom(Walker);
4021 if (ParentList[OtherAtom->nr] != NULL) { // if otheratom is also a father of an atom on this molecule, create the bond
4022 *out << Verbose(4) << "Endpoints of Bond " << Father->ListOfBondsPerAtom[Walker->nr][i] << " are both present: " << ParentList[Walker->nr]->Name << " and " << ParentList[OtherAtom->nr]->Name << "." << endl;
4023 AddBond(ParentList[Walker->nr], ParentList[OtherAtom->nr], Father->ListOfBondsPerAtom[Walker->nr][i]->BondDegree);
4024 }
4025 }
4026 }
4027 }
4028 }
4029
4030 Free((void **)&ParentList, "molecule::BuildInducedSubgraph: **ParentList");
4031 *out << Verbose(2) << "End of BuildInducedSubgraph." << endl;
4032 return status;
4033};
4034
4035
4036/** Looks through a StackClass<atom *> and returns the likeliest removal candiate.
4037 * \param *out output stream for debugging messages
4038 * \param *&Leaf KeySet to look through
4039 * \param *&ShortestPathList list of the shortest path to decide which atom to suggest as removal candidate in the end
4040 * \param index of the atom suggested for removal
4041 */
4042int molecule::LookForRemovalCandidate(ofstream *&out, KeySet *&Leaf, int *&ShortestPathList)
4043{
4044 atom *Runner = NULL;
4045 int SP, Removal;
4046
4047 *out << Verbose(2) << "Looking for removal candidate." << endl;
4048 SP = -1; //0; // not -1, so that Root is never removed
4049 Removal = -1;
4050 for (KeySet::iterator runner = Leaf->begin(); runner != Leaf->end(); runner++) {
4051 Runner = FindAtom((*runner));
4052 if (Runner->type->Z != 1) { // skip all those added hydrogens when re-filling snake stack
4053 if (ShortestPathList[(*runner)] > SP) { // remove the oldest one with longest shortest path
4054 SP = ShortestPathList[(*runner)];
4055 Removal = (*runner);
4056 }
4057 }
4058 }
4059 return Removal;
4060};
4061
4062/** Stores a fragment from \a KeySet into \a molecule.
4063 * First creates the minimal set of atoms from the KeySet, then creates the bond structure from the complete
4064 * molecule and adds missing hydrogen where bonds were cut.
4065 * \param *out output stream for debugging messages
4066 * \param &Leaflet pointer to KeySet structure
4067 * \param IsAngstroem whether we have Ansgtroem or bohrradius
4068 * \return pointer to constructed molecule
4069 */
4070molecule * molecule::StoreFragmentFromKeySet(ofstream *out, KeySet &Leaflet, bool IsAngstroem)
4071{
4072 atom *Runner = NULL, *FatherOfRunner = NULL, *OtherFather = NULL;
4073 atom **SonList = (atom **) Malloc(sizeof(atom *)*AtomCount, "molecule::StoreFragmentFromStack: **SonList");
4074 molecule *Leaf = new molecule(elemente);
4075 bool LonelyFlag = false;
4076 int size;
4077
4078// *out << Verbose(1) << "Begin of StoreFragmentFromKeyset." << endl;
4079
4080 Leaf->BondDistance = BondDistance;
4081 for(int i=NDIM*2;i--;)
4082 Leaf->cell_size[i] = cell_size[i];
4083
4084 // initialise SonList (indicates when we need to replace a bond with hydrogen instead)
4085 for(int i=AtomCount;i--;)
4086 SonList[i] = NULL;
4087
4088 // first create the minimal set of atoms from the KeySet
4089 size = 0;
4090 for(KeySet::iterator runner = Leaflet.begin(); runner != Leaflet.end(); runner++) {
4091 FatherOfRunner = FindAtom((*runner)); // find the id
4092 SonList[FatherOfRunner->nr] = Leaf->AddCopyAtom(FatherOfRunner);
4093 size++;
4094 }
4095
4096 // create the bonds between all: Make it an induced subgraph and add hydrogen
4097// *out << Verbose(2) << "Creating bonds from father graph (i.e. induced subgraph creation)." << endl;
4098 Runner = Leaf->start;
4099 while (Runner->next != Leaf->end) {
4100 Runner = Runner->next;
4101 LonelyFlag = true;
4102 FatherOfRunner = Runner->father;
4103 if (SonList[FatherOfRunner->nr] != NULL) { // check if this, our father, is present in list
4104 // create all bonds
4105 for (int i=0;i<NumberOfBondsPerAtom[FatherOfRunner->nr];i++) { // go through every bond of father
4106 OtherFather = ListOfBondsPerAtom[FatherOfRunner->nr][i]->GetOtherAtom(FatherOfRunner);
4107// *out << Verbose(2) << "Father " << *FatherOfRunner << " of son " << *SonList[FatherOfRunner->nr] << " is bound to " << *OtherFather;
4108 if (SonList[OtherFather->nr] != NULL) {
4109// *out << ", whose son is " << *SonList[OtherFather->nr] << "." << endl;
4110 if (OtherFather->nr > FatherOfRunner->nr) { // add bond (nr check is for adding only one of both variants: ab, ba)
4111// *out << Verbose(3) << "Adding Bond: ";
4112// *out <<
4113 Leaf->AddBond(Runner, SonList[OtherFather->nr], ListOfBondsPerAtom[FatherOfRunner->nr][i]->BondDegree);
4114// *out << "." << endl;
4115 //NumBonds[Runner->nr]++;
4116 } else {
4117// *out << Verbose(3) << "Not adding bond, labels in wrong order." << endl;
4118 }
4119 LonelyFlag = false;
4120 } else {
4121// *out << ", who has no son in this fragment molecule." << endl;
4122#ifdef ADDHYDROGEN
4123 //*out << Verbose(3) << "Adding Hydrogen to " << Runner->Name << " and a bond in between." << endl;
4124 Leaf->AddHydrogenReplacementAtom(out, ListOfBondsPerAtom[FatherOfRunner->nr][i], Runner, FatherOfRunner, OtherFather, ListOfBondsPerAtom[FatherOfRunner->nr],NumberOfBondsPerAtom[FatherOfRunner->nr], IsAngstroem);
4125#endif
4126 //NumBonds[Runner->nr] += ListOfBondsPerAtom[FatherOfRunner->nr][i]->BondDegree;
4127 }
4128 }
4129 } else {
4130 *out << Verbose(0) << "ERROR: Son " << Runner->Name << " has father " << FatherOfRunner->Name << " but its entry in SonList is " << SonList[FatherOfRunner->nr] << "!" << endl;
4131 }
4132 if ((LonelyFlag) && (size > 1)) {
4133 *out << Verbose(0) << *Runner << "has got bonds only to hydrogens!" << endl;
4134 }
4135#ifdef ADDHYDROGEN
4136 while ((Runner->next != Leaf->end) && (Runner->next->type->Z == 1)) // skip added hydrogen
4137 Runner = Runner->next;
4138#endif
4139 }
4140 Leaf->CreateListOfBondsPerAtom(out);
4141 //Leaflet->Leaf->ScanForPeriodicCorrection(out);
4142 Free((void **)&SonList, "molecule::StoreFragmentFromStack: **SonList");
4143// *out << Verbose(1) << "End of StoreFragmentFromKeyset." << endl;
4144 return Leaf;
4145};
4146
4147/** Creates \a MoleculeListClass of all unique fragments of the \a molecule containing \a Order atoms or vertices.
4148 * The picture to have in mind is that of a DFS "snake" of a certain length \a Order, i.e. as in the infamous
4149 * computer game, that winds through the connected graph representing the molecule. Color (white,
4150 * lightgray, darkgray, black) indicates whether a vertex has been discovered so far or not. Labels will help in
4151 * creating only unique fragments and not additional ones with vertices simply in different sequence.
4152 * The Predecessor is always the one that came before in discovering, needed on backstepping. And
4153 * finally, the ShortestPath is needed for removing vertices from the snake stack during the back-
4154 * stepping.
4155 * \param *out output stream for debugging
4156 * \param Order number of atoms in each fragment
4157 * \param *configuration configuration for writing config files for each fragment
4158 * \return List of all unique fragments with \a Order atoms
4159 */
4160/*
4161MoleculeListClass * molecule::CreateListOfUniqueFragmentsOfOrder(ofstream *out, int Order, config *configuration)
4162{
4163 atom **PredecessorList = (atom **) Malloc(sizeof(atom *)*AtomCount, "molecule::CreateListOfUniqueFragmentsOfOrder: **PredecessorList");
4164 int *ShortestPathList = (int *) Malloc(sizeof(int)*AtomCount, "molecule::CreateListOfUniqueFragmentsOfOrder: *ShortestPathList");
4165 int *Labels = (int *) Malloc(sizeof(int)*AtomCount, "molecule::CreateListOfUniqueFragmentsOfOrder: *Labels");
4166 enum Shading *ColorVertexList = (enum Shading *) Malloc(sizeof(enum Shading)*AtomCount, "molecule::CreateListOfUniqueFragmentsOfOrder: *ColorList");
4167 enum Shading *ColorEdgeList = (enum Shading *) Malloc(sizeof(enum Shading)*BondCount, "molecule::CreateListOfUniqueFragmentsOfOrder: *ColorBondList");
4168 StackClass<atom *> *RootStack = new StackClass<atom *>(AtomCount);
4169 StackClass<atom *> *TouchedStack = new StackClass<atom *>((int)pow(4,Order)+2); // number of atoms reached from one with maximal 4 bonds plus Root itself
4170 StackClass<atom *> *SnakeStack = new StackClass<atom *>(Order+1); // equal to Order is not possible, as then the StackClass<atom *> cannot discern between full and empty stack!
4171 MoleculeLeafClass *Leaflet = NULL, *TempLeaf = NULL;
4172 MoleculeListClass *FragmentList = NULL;
4173 atom *Walker = NULL, *OtherAtom = NULL, *Root = NULL, *Removal = NULL;
4174 bond *Binder = NULL;
4175 int RunningIndex = 0, FragmentCounter = 0;
4176
4177 *out << Verbose(1) << "Begin of CreateListOfUniqueFragmentsOfOrder." << endl;
4178
4179 // reset parent list
4180 *out << Verbose(3) << "Resetting labels, parent, predecessor, color and shortest path lists." << endl;
4181 for (int i=0;i<AtomCount;i++) { // reset all atom labels
4182 // initialise each vertex as white with no predecessor, empty queue, color lightgray, not labelled, no sons
4183 Labels[i] = -1;
4184 SonList[i] = NULL;
4185 PredecessorList[i] = NULL;
4186 ColorVertexList[i] = white;
4187 ShortestPathList[i] = -1;
4188 }
4189 for (int i=0;i<BondCount;i++)
4190 ColorEdgeList[i] = white;
4191 RootStack->ClearStack(); // clearstack and push first atom if exists
4192 TouchedStack->ClearStack();
4193 Walker = start->next;
4194 while ((Walker != end)
4195#ifdef ADDHYDROGEN
4196 && (Walker->type->Z == 1)
4197#endif
4198 ) { // search for first non-hydrogen atom
4199 *out << Verbose(4) << "Current Root candidate is " << Walker->Name << "." << endl;
4200 Walker = Walker->next;
4201 }
4202 if (Walker != end)
4203 RootStack->Push(Walker);
4204 else
4205 *out << Verbose(0) << "ERROR: Could not find an appropriate Root atom!" << endl;
4206 *out << Verbose(3) << "Root " << Walker->Name << " is on AtomStack, beginning loop through all vertices ..." << endl;
4207
4208 ///// OUTER LOOP ////////////
4209 while (!RootStack->IsEmpty()) {
4210 // get new root vertex from atom stack
4211 Root = RootStack->PopFirst();
4212 ShortestPathList[Root->nr] = 0;
4213 if (Labels[Root->nr] == -1)
4214 Labels[Root->nr] = RunningIndex++; // prevent it from getting again on AtomStack
4215 PredecessorList[Root->nr] = Root;
4216 TouchedStack->Push(Root);
4217 *out << Verbose(0) << "Root for this loop is: " << Root->Name << ".\n";
4218
4219 // clear snake stack
4220 SnakeStack->ClearStack();
4221 //SnakeStack->TestImplementation(out, start->next);
4222
4223 ///// INNER LOOP ////////////
4224 // Problems:
4225 // - what about cyclic bonds?
4226 Walker = Root;
4227 do {
4228 *out << Verbose(1) << "Current Walker is: " << Walker->Name;
4229 // initial setting of the new Walker: label, color, shortest path and put on stacks
4230 if (Labels[Walker->nr] == -1) { // give atom a unique, monotonely increasing number
4231 Labels[Walker->nr] = RunningIndex++;
4232 RootStack->Push(Walker);
4233 }
4234 *out << ", has label " << Labels[Walker->nr];
4235 if ((ColorVertexList[Walker->nr] == white) || ((Binder != NULL) && (ColorEdgeList[Binder->nr] == white))) { // color it if newly discovered and push on stacks (and if within reach!)
4236 if ((Binder != NULL) && (ColorEdgeList[Binder->nr] == white)) {
4237 // Binder ought to be set still from last neighbour search
4238 *out << ", coloring bond " << *Binder << " black";
4239 ColorEdgeList[Binder->nr] = black; // mark this bond as used
4240 }
4241 if (ShortestPathList[Walker->nr] == -1) {
4242 ShortestPathList[Walker->nr] = ShortestPathList[PredecessorList[Walker->nr]->nr]+1;
4243 TouchedStack->Push(Walker); // mark every atom for lists cleanup later, whose shortest path has been changed
4244 }
4245 if ((ShortestPathList[Walker->nr] < Order) && (ColorVertexList[Walker->nr] != darkgray)) { // if not already on snake stack
4246 SnakeStack->Push(Walker);
4247 ColorVertexList[Walker->nr] = darkgray; // mark as dark gray of on snake stack
4248 }
4249 }
4250 *out << ", SP of " << ShortestPathList[Walker->nr] << " and its color is " << GetColor(ColorVertexList[Walker->nr]) << "." << endl;
4251
4252 // then check the stack for a newly stumbled upon fragment
4253 if (SnakeStack->ItemCount() == Order) { // is stack full?
4254 // store the fragment if it is one and get a removal candidate
4255 Removal = StoreFragmentFromStack(out, Root, Walker, Leaflet, SnakeStack, ShortestPathList, SonList, Labels, &FragmentCounter, configuration);
4256 // remove the candidate if one was found
4257 if (Removal != NULL) {
4258 *out << Verbose(2) << "Removing item " << Removal->Name << " with SP of " << ShortestPathList[Removal->nr] << " from snake stack." << endl;
4259 SnakeStack->RemoveItem(Removal);
4260 ColorVertexList[Removal->nr] = lightgray; // return back to not on snake stack but explored marking
4261 if (Walker == Removal) { // if the current atom is to be removed, we also have to take a step back
4262 Walker = PredecessorList[Removal->nr];
4263 *out << Verbose(2) << "Stepping back to " << Walker->Name << "." << endl;
4264 }
4265 }
4266 } else
4267 Removal = NULL;
4268
4269 // finally, look for a white neighbour as the next Walker
4270 Binder = NULL;
4271 if ((Removal == NULL) || (Walker != PredecessorList[Removal->nr])) { // don't look, if a new walker has been set above
4272 *out << Verbose(2) << "Snake has currently " << SnakeStack->ItemCount() << " item(s)." << endl;
4273 OtherAtom = NULL; // this is actually not needed, every atom has at least one neighbour
4274 if (ShortestPathList[Walker->nr] < Order) {
4275 for(int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++) {
4276 Binder = ListOfBondsPerAtom[Walker->nr][i];
4277 *out << Verbose(2) << "Current bond is " << *Binder << ": ";
4278 OtherAtom = Binder->GetOtherAtom(Walker);
4279 if ((Labels[OtherAtom->nr] != -1) && (Labels[OtherAtom->nr] < Labels[Root->nr])) { // we don't step up to labels bigger than us
4280 *out << "Label " << Labels[OtherAtom->nr] << " is smaller than Root's " << Labels[Root->nr] << "." << endl;
4281 //ColorVertexList[OtherAtom->nr] = lightgray; // mark as explored
4282 } else { // otherwise check its colour and element
4283 if (
4284#ifdef ADDHYDROGEN
4285 (OtherAtom->type->Z != 1) &&
4286#endif
4287 (ColorEdgeList[Binder->nr] == white)) { // skip hydrogen, look for unexplored vertices
4288 *out << "Moving along " << GetColor(ColorEdgeList[Binder->nr]) << " bond " << Binder << " to " << ((ColorVertexList[OtherAtom->nr] == white) ? "unexplored" : "explored") << " item: " << OtherAtom->Name << "." << endl;
4289 // i find it currently rather sensible to always set the predecessor in order to find one's way back
4290 //if (PredecessorList[OtherAtom->nr] == NULL) {
4291 PredecessorList[OtherAtom->nr] = Walker;
4292 *out << Verbose(3) << "Setting Predecessor of " << OtherAtom->Name << " to " << PredecessorList[OtherAtom->nr]->Name << "." << endl;
4293 //} else {
4294 // *out << Verbose(3) << "Predecessor of " << OtherAtom->Name << " is " << PredecessorList[OtherAtom->nr]->Name << "." << endl;
4295 //}
4296 Walker = OtherAtom;
4297 break;
4298 } else {
4299 if (OtherAtom->type->Z == 1)
4300 *out << "Links to a hydrogen atom." << endl;
4301 else
4302 *out << "Bond has not white but " << GetColor(ColorEdgeList[Binder->nr]) << " color." << endl;
4303 }
4304 }
4305 }
4306 } else { // means we have stepped beyond the horizon: Return!
4307 Walker = PredecessorList[Walker->nr];
4308 OtherAtom = Walker;
4309 *out << Verbose(3) << "We have gone too far, stepping back to " << Walker->Name << "." << endl;
4310 }
4311 if (Walker != OtherAtom) { // if no white neighbours anymore, color it black
4312 *out << Verbose(2) << "Coloring " << Walker->Name << " black." << endl;
4313 ColorVertexList[Walker->nr] = black;
4314 Walker = PredecessorList[Walker->nr];
4315 }
4316 }
4317 } while ((Walker != Root) || (ColorVertexList[Root->nr] != black));
4318 *out << Verbose(2) << "Inner Looping is finished." << endl;
4319
4320 // if we reset all AtomCount atoms, we have again technically O(N^2) ...
4321 *out << Verbose(2) << "Resetting lists." << endl;
4322 Walker = NULL;
4323 Binder = NULL;
4324 while (!TouchedStack->IsEmpty()) {
4325 Walker = TouchedStack->PopLast();
4326 *out << Verbose(3) << "Re-initialising entries of " << *Walker << "." << endl;
4327 for(int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++)
4328 ColorEdgeList[ListOfBondsPerAtom[Walker->nr][i]->nr] = white;
4329 PredecessorList[Walker->nr] = NULL;
4330 ColorVertexList[Walker->nr] = white;
4331 ShortestPathList[Walker->nr] = -1;
4332 }
4333 }
4334 *out << Verbose(1) << "Outer Looping over all vertices is done." << endl;
4335
4336 // copy together
4337 *out << Verbose(1) << "Copying all fragments into MoleculeList structure." << endl;
4338 FragmentList = new MoleculeListClass(FragmentCounter, AtomCount);
4339 RunningIndex = 0;
4340 while ((Leaflet != NULL) && (RunningIndex < FragmentCounter)) {
4341 FragmentList->ListOfMolecules[RunningIndex++] = Leaflet->Leaf;
4342 Leaflet->Leaf = NULL; // prevent molecule from being removed
4343 TempLeaf = Leaflet;
4344 Leaflet = Leaflet->previous;
4345 delete(TempLeaf);
4346 };
4347
4348 // free memory and exit
4349 Free((void **)&PredecessorList, "molecule::CreateListOfUniqueFragmentsOfOrder: **PredecessorList");
4350 Free((void **)&ShortestPathList, "molecule::CreateListOfUniqueFragmentsOfOrder: *ShortestPathList");
4351 Free((void **)&Labels, "molecule::CreateListOfUniqueFragmentsOfOrder: *Labels");
4352 Free((void **)&ColorVertexList, "molecule::CreateListOfUniqueFragmentsOfOrder: *ColorList");
4353 delete(RootStack);
4354 delete(TouchedStack);
4355 delete(SnakeStack);
4356
4357 *out << Verbose(1) << "End of CreateListOfUniqueFragmentsOfOrder." << endl;
4358 return FragmentList;
4359};
4360*/
4361
4362/** Structure containing all values in power set combination generation.
4363 */
4364struct UniqueFragments {
4365 config *configuration;
4366 atom *Root;
4367 Graph *Leaflet;
4368 KeySet *FragmentSet;
4369 int ANOVAOrder;
4370 int FragmentCounter;
4371 int CurrentIndex;
4372 double TEFactor;
4373 int *ShortestPathList;
4374 bool **UsedList;
4375 bond **BondsPerSPList;
4376 int *BondsPerSPCount;
4377};
4378
4379/** From a given set of Bond sorted by Shortest Path distance, create all possible fragments of size \a SetDimension.
4380 * -# loops over every possible combination (2^dimension of edge set)
4381 * -# inserts current set, if there's still space left
4382 * -# yes: calls SPFragmentGenerator with structure, created new edge list and size respective to root dist
4383ance+1
4384 * -# no: stores fragment into keyset list by calling InsertFragmentIntoGraph
4385 * -# removes all items added into the snake stack (in UniqueFragments structure) added during level (root
4386distance) and current set
4387 * \param *out output stream for debugging
4388 * \param FragmentSearch UniqueFragments structure with all values needed
4389 * \param RootDistance current shortest path level, whose set of edges is represented by **BondsSet
4390 * \param SetDimension Number of possible bonds on this level (i.e. size of the array BondsSet[])
4391 * \param SubOrder remaining number of allowed vertices to add
4392 */
4393void molecule::SPFragmentGenerator(ofstream *out, struct UniqueFragments *FragmentSearch, int RootDistance, bond **BondsSet, int SetDimension, int SubOrder)
4394{
4395 atom *OtherWalker = NULL;
4396 int verbosity = 0; //FragmentSearch->ANOVAOrder-SubOrder;
4397 int NumCombinations;
4398 bool bit;
4399 int bits, TouchedIndex, SubSetDimension, SP, Added;
4400 int Removal;
4401 int SpaceLeft;
4402 int *TouchedList = (int *) Malloc(sizeof(int)*(SubOrder+1), "molecule::SPFragmentGenerator: *TouchedList");
4403 bond *Binder = NULL;
4404 bond **BondsList = NULL;
4405 KeySetTestPair TestKeySetInsert;
4406
4407 NumCombinations = 1 << SetDimension;
4408
4409 // Hier muessen von 1 bis NumberOfBondsPerAtom[Walker->nr] alle Kombinationen
4410 // von Endstuecken (aus den Bonds) hinzugefᅵᅵgt werden und fᅵᅵr verbleibende ANOVAOrder
4411 // rekursiv GraphCrawler in der nᅵᅵchsten Ebene aufgerufen werden
4412
4413 *out << Verbose(1+verbosity) << "Begin of SPFragmentGenerator." << endl;
4414 *out << Verbose(1+verbosity) << "We are " << RootDistance << " away from Root, which is " << *FragmentSearch->Root << ", SubOrder is " << SubOrder << ", SetDimension is " << SetDimension << " and this means " << NumCombinations-1 << " combination(s)." << endl;
4415
4416 // initialised touched list (stores added atoms on this level)
4417 *out << Verbose(1+verbosity) << "Clearing touched list." << endl;
4418 for (TouchedIndex=SubOrder+1;TouchedIndex--;) // empty touched list
4419 TouchedList[TouchedIndex] = -1;
4420 TouchedIndex = 0;
4421
4422 // create every possible combination of the endpieces
4423 *out << Verbose(1+verbosity) << "Going through all combinations of the power set." << endl;
4424 for (int i=1;i<NumCombinations;i++) { // sweep through all power set combinations (skip empty set!)
4425 // count the set bit of i
4426 bits = 0;
4427 for (int j=SetDimension;j--;)
4428 bits += (i & (1 << j)) >> j;
4429
4430 *out << Verbose(1+verbosity) << "Current set is " << Binary(i | (1 << SetDimension)) << ", number of bits is " << bits << "." << endl;
4431 if (bits <= SubOrder) { // if not greater than additional atoms allowed on stack, continue
4432 // --1-- add this set of the power set of bond partners to the snake stack
4433 Added = 0;
4434 for (int j=0;j<SetDimension;j++) { // pull out every bit by shifting
4435 bit = ((i & (1 << j)) != 0); // mask the bit for the j-th bond
4436 if (bit) { // if bit is set, we add this bond partner
4437 OtherWalker = BondsSet[j]->rightatom; // rightatom is always the one more distant, i.e. the one to add
4438 //*out << Verbose(1+verbosity) << "Current Bond is " << ListOfBondsPerAtom[Walker->nr][i] << ", checking on " << *OtherWalker << "." << endl;
4439 *out << Verbose(2+verbosity) << "Adding " << *OtherWalker << " with nr " << OtherWalker->nr << "." << endl;
4440 TestKeySetInsert = FragmentSearch->FragmentSet->insert(OtherWalker->nr);
4441 if (TestKeySetInsert.second) {
4442 TouchedList[TouchedIndex++] = OtherWalker->nr; // note as added
4443 Added++;
4444 } else {
4445 *out << Verbose(2+verbosity) << "This was item was already present in the keyset." << endl;
4446 }
4447 //FragmentSearch->UsedList[OtherWalker->nr][i] = true;
4448 //}
4449 } else {
4450 *out << Verbose(2+verbosity) << "Not adding." << endl;
4451 }
4452 }
4453
4454 SpaceLeft = SubOrder - Added ;// SubOrder - bits; // due to item's maybe being already present, this does not work anymore
4455 if (SpaceLeft > 0) {
4456 *out << Verbose(1+verbosity) << "There's still some space left on stack: " << SpaceLeft << "." << endl;
4457 if (SubOrder > 1) { // Due to Added above we have to check extra whether we're not already reaching beyond the desired Order
4458 // --2-- look at all added end pieces of this combination, construct bond subsets and sweep through a power set of these by recursion
4459 SP = RootDistance+1; // this is the next level
4460 // first count the members in the subset
4461 SubSetDimension = 0;
4462 Binder = FragmentSearch->BondsPerSPList[2*SP]; // start node for this level
4463 while (Binder->next != FragmentSearch->BondsPerSPList[2*SP+1]) { // compare to end node of this level
4464 Binder = Binder->next;
4465 for (int k=TouchedIndex;k--;) {
4466 if (Binder->Contains(TouchedList[k])) // if we added this very endpiece
4467 SubSetDimension++;
4468 }
4469 }
4470 // then allocate and fill the list
4471 BondsList = (bond **) Malloc(sizeof(bond *)*SubSetDimension, "molecule::SPFragmentGenerator: **BondsList");
4472 SubSetDimension = 0;
4473 Binder = FragmentSearch->BondsPerSPList[2*SP];
4474 while (Binder->next != FragmentSearch->BondsPerSPList[2*SP+1]) {
4475 Binder = Binder->next;
4476 for (int k=0;k<TouchedIndex;k++) {
4477 if (Binder->leftatom->nr == TouchedList[k]) // leftatom is always the close one
4478 BondsList[SubSetDimension++] = Binder;
4479 }
4480 }
4481 *out << Verbose(2+verbosity) << "Calling subset generator " << SP << " away from root " << *FragmentSearch->Root << " with sub set dimension " << SubSetDimension << "." << endl;
4482 SPFragmentGenerator(out, FragmentSearch, SP, BondsList, SubSetDimension, SubOrder-bits);
4483 Free((void **)&BondsList, "molecule::SPFragmentGenerator: **BondsList");
4484 }
4485 } else {
4486 // --2-- otherwise store the complete fragment
4487 *out << Verbose(1+verbosity) << "Enough items on stack for a fragment!" << endl;
4488 // store fragment as a KeySet
4489 *out << Verbose(2) << "Found a new fragment[" << FragmentSearch->FragmentCounter << "], local nr.s are: ";
4490 for(KeySet::iterator runner = FragmentSearch->FragmentSet->begin(); runner != FragmentSearch->FragmentSet->end(); runner++)
4491 *out << (*runner) << " ";
4492 *out << endl;
4493 //if (!CheckForConnectedSubgraph(out, FragmentSearch->FragmentSet))
4494 //*out << Verbose(0) << "ERROR: The found fragment is not a connected subgraph!" << endl;
4495 InsertFragmentIntoGraph(out, FragmentSearch);
4496 //Removal = LookForRemovalCandidate(out, FragmentSearch->FragmentSet, FragmentSearch->ShortestPathList);
4497 //Removal = StoreFragmentFromStack(out, FragmentSearch->Root, FragmentSearch->Leaflet, FragmentSearch->FragmentStack, FragmentSearch->ShortestPathList, &FragmentSearch->FragmentCounter, FragmentSearch->configuration);
4498 }
4499
4500 // --3-- remove all added items in this level from snake stack
4501 *out << Verbose(1+verbosity) << "Removing all items that were added on this SP level " << RootDistance << "." << endl;
4502 for(int j=0;j<TouchedIndex;j++) {
4503 Removal = TouchedList[j];
4504 *out << Verbose(2+verbosity) << "Removing item nr. " << Removal << " from snake stack." << endl;
4505 FragmentSearch->FragmentSet->erase(Removal);
4506 TouchedList[j] = -1;
4507 }
4508 *out << Verbose(2) << "Remaining local nr.s on snake stack are: ";
4509 for(KeySet::iterator runner = FragmentSearch->FragmentSet->begin(); runner != FragmentSearch->FragmentSet->end(); runner++)
4510 *out << (*runner) << " ";
4511 *out << endl;
4512 TouchedIndex = 0; // set Index to 0 for list of atoms added on this level
4513 } else {
4514 *out << Verbose(2+verbosity) << "More atoms to add for this set (" << bits << ") than space left on stack " << SubOrder << ", skipping this set." << endl;
4515 }
4516 }
4517 Free((void **)&TouchedList, "molecule::SPFragmentGenerator: *TouchedList");
4518 *out << Verbose(1+verbosity) << "End of SPFragmentGenerator, " << RootDistance << " away from Root " << *FragmentSearch->Root << " and SubOrder is " << SubOrder << "." << endl;
4519};
4520
4521/** For a given keyset \a *Fragment, checks whether it is connected in the current molecule.
4522 * \param *out output stream for debugging
4523 * \param *Fragment Keyset of fragment's vertices
4524 * \return true - connected, false - disconnected
4525 * \note this is O(n^2) for it's just a bug checker not meant for permanent use!
4526 */
4527bool molecule::CheckForConnectedSubgraph(ofstream *out, KeySet *Fragment)
4528{
4529 atom *Walker = NULL, *Walker2 = NULL;
4530 bool BondStatus = false;
4531 int size;
4532
4533 *out << Verbose(1) << "Begin of CheckForConnectedSubgraph" << endl;
4534 *out << Verbose(2) << "Disconnected atom: ";
4535
4536 // count number of atoms in graph
4537 size = 0;
4538 for(KeySet::iterator runner = Fragment->begin(); runner != Fragment->end(); runner++)
4539 size++;
4540 if (size > 1)
4541 for(KeySet::iterator runner = Fragment->begin(); runner != Fragment->end(); runner++) {
4542 Walker = FindAtom(*runner);
4543 BondStatus = false;
4544 for(KeySet::iterator runners = Fragment->begin(); runners != Fragment->end(); runners++) {
4545 Walker2 = FindAtom(*runners);
4546 for (int i=0;i<NumberOfBondsPerAtom[Walker->nr]; i++) {
4547 if (ListOfBondsPerAtom[Walker->nr][i]->GetOtherAtom(Walker) == Walker2) {
4548 BondStatus = true;
4549 break;
4550 }
4551 if (BondStatus)
4552 break;
4553 }
4554 }
4555 if (!BondStatus) {
4556 *out << (*Walker) << endl;
4557 return false;
4558 }
4559 }
4560 else {
4561 *out << "none." << endl;
4562 return true;
4563 }
4564 *out << "none." << endl;
4565
4566 *out << Verbose(1) << "End of CheckForConnectedSubgraph" << endl;
4567
4568 return true;
4569}
4570
4571/** Creates a list of all unique fragments of certain vertex size from a given graph \a Fragment for a given root vertex in the context of \a this molecule.
4572 * -# initialises UniqueFragments structure
4573 * -# fills edge list via BFS
4574 * -# creates the fragment by calling recursive function SPFragmentGenerator with UniqueFragments structure, 0 as
4575 root distance, the edge set, its dimension and the current suborder
4576 * -# Free'ing structure
4577 * Note that we may use the fact that the atoms are SP-ordered on the atomstack. I.e. when popping always the last, we first get all
4578 * with SP of 2, then those with SP of 3, then those with SP of 4 and so on.
4579 * \param *out output stream for debugging
4580 * \param Order bond order (limits BFS exploration and "number of digits" in power set generation
4581 * \param FragmentSearch UniqueFragments structure containing TEFactor, root atom and so on
4582 * \param RestrictedKeySet Restricted vertex set to use in context of molecule
4583 * \return number of inserted fragments
4584 * \note ShortestPathList in FragmentSearch structure is probably due to NumberOfAtomsSPLevel and SP not needed anymore
4585 */
4586int molecule::PowerSetGenerator(ofstream *out, int Order, struct UniqueFragments &FragmentSearch, KeySet RestrictedKeySet)
4587{
4588 int SP, AtomKeyNr;
4589 atom *Walker = NULL, *OtherWalker = NULL, *Predecessor = NULL;
4590 bond *Binder = NULL;
4591 bond *CurrentEdge = NULL;
4592 bond **BondsList = NULL;
4593 int RootKeyNr = FragmentSearch.Root->GetTrueFather()->nr;
4594 int Counter = FragmentSearch.FragmentCounter;
4595 int RemainingWalkers;
4596
4597 *out << endl;
4598 *out << Verbose(0) << "Begin of PowerSetGenerator with order " << Order << " at Root " << *FragmentSearch.Root << "." << endl;
4599
4600 // prepare Label and SP arrays of the BFS search
4601 FragmentSearch.ShortestPathList[FragmentSearch.Root->nr] = 0;
4602
4603 // prepare root level (SP = 0) and a loop bond denoting Root
4604 for (int i=1;i<Order;i++)
4605 FragmentSearch.BondsPerSPCount[i] = 0;
4606 FragmentSearch.BondsPerSPCount[0] = 1;
4607 Binder = new bond(FragmentSearch.Root, FragmentSearch.Root);
4608 add(Binder, FragmentSearch.BondsPerSPList[1]);
4609
4610 // do a BFS search to fill the SP lists and label the found vertices
4611 // Actually, we should construct a spanning tree vom the root atom and select all edges therefrom and put them into
4612 // according shortest path lists. However, we don't. Rather we fill these lists right away, as they do form a spanning
4613 // tree already sorted into various SP levels. That's why we just do loops over the depth (CurrentSP) and breadth
4614 // (EdgeinSPLevel) of this tree ...
4615 // In another picture, the bonds always contain a direction by rightatom being the one more distant from root and hence
4616 // naturally leftatom forming its predecessor, preventing the BFS"seeker" from continuing in the wrong direction.
4617 *out << endl;
4618 *out << Verbose(0) << "Starting BFS analysis ..." << endl;
4619 for (SP = 0; SP < (Order-1); SP++) {
4620 *out << Verbose(1) << "New SP level reached: " << SP << ", creating new SP list with " << FragmentSearch.BondsPerSPCount[SP] << " item(s)";
4621 if (SP > 0) {
4622 *out << ", old level closed with " << FragmentSearch.BondsPerSPCount[SP-1] << " item(s)." << endl;
4623 FragmentSearch.BondsPerSPCount[SP] = 0;
4624 } else
4625 *out << "." << endl;
4626
4627 RemainingWalkers = FragmentSearch.BondsPerSPCount[SP];
4628 CurrentEdge = FragmentSearch.BondsPerSPList[2*SP]; /// start of this SP level's list
4629 while (CurrentEdge->next != FragmentSearch.BondsPerSPList[2*SP+1]) { /// end of this SP level's list
4630 CurrentEdge = CurrentEdge->next;
4631 RemainingWalkers--;
4632 Walker = CurrentEdge->rightatom; // rightatom is always the one more distant
4633 Predecessor = CurrentEdge->leftatom; // ... and leftatom is predecessor
4634 AtomKeyNr = Walker->nr;
4635 *out << Verbose(0) << "Current Walker is: " << *Walker << " with nr " << Walker->nr << " and SP of " << SP << ", with " << RemainingWalkers << " remaining walkers on this level." << endl;
4636 // check for new sp level
4637 // go through all its bonds
4638 *out << Verbose(1) << "Going through all bonds of Walker." << endl;
4639 for (int i=0;i<NumberOfBondsPerAtom[AtomKeyNr];i++) {
4640 Binder = ListOfBondsPerAtom[AtomKeyNr][i];
4641 OtherWalker = Binder->GetOtherAtom(Walker);
4642 if ((RestrictedKeySet.find(OtherWalker->nr) != RestrictedKeySet.end())
4643 #ifdef ADDHYDROGEN
4644 && (OtherWalker->type->Z != 1)
4645 #endif
4646 ) { // skip hydrogens and restrict to fragment
4647 *out << Verbose(2) << "Current partner is " << *OtherWalker << " with nr " << OtherWalker->nr << " in bond " << *Binder << "." << endl;
4648 // set the label if not set (and push on root stack as well)
4649 if ((OtherWalker != Predecessor) && (OtherWalker->GetTrueFather()->nr > RootKeyNr)) { // only pass through those with label bigger than Root's
4650 FragmentSearch.ShortestPathList[OtherWalker->nr] = SP+1;
4651 *out << Verbose(3) << "Set Shortest Path to " << FragmentSearch.ShortestPathList[OtherWalker->nr] << "." << endl;
4652 // add the bond in between to the SP list
4653 Binder = new bond(Walker, OtherWalker); // create a new bond in such a manner, that bond::rightatom is always the one more distant
4654 add(Binder, FragmentSearch.BondsPerSPList[2*(SP+1)+1]);
4655 FragmentSearch.BondsPerSPCount[SP+1]++;
4656 *out << Verbose(3) << "Added its bond to SP list, having now " << FragmentSearch.BondsPerSPCount[SP+1] << " item(s)." << endl;
4657 } else {
4658 if (OtherWalker != Predecessor)
4659 *out << Verbose(3) << "Not passing on, as index of " << *OtherWalker << " " << OtherWalker->GetTrueFather()->nr << " is smaller than that of Root " << RootKeyNr << "." << endl;
4660 else
4661 *out << Verbose(3) << "This is my predecessor " << *Predecessor << "." << endl;
4662 }
4663 } else *out << Verbose(2) << "Is not in the restricted keyset or skipping hydrogen " << *OtherWalker << "." << endl;
4664 }
4665 }
4666 }
4667
4668 // outputting all list for debugging
4669 *out << Verbose(0) << "Printing all found lists." << endl;
4670 for(int i=1;i<Order;i++) { // skip the root edge in the printing
4671 Binder = FragmentSearch.BondsPerSPList[2*i];
4672 *out << Verbose(1) << "Current SP level is " << i << "." << endl;
4673 while (Binder->next != FragmentSearch.BondsPerSPList[2*i+1]) {
4674 Binder = Binder->next;
4675 *out << Verbose(2) << *Binder << endl;
4676 }
4677 }
4678
4679 // creating fragments with the found edge sets (may be done in reverse order, faster)
4680 SP = -1; // the Root <-> Root edge must be subtracted!
4681 for(int i=Order;i--;) { // sum up all found edges
4682 Binder = FragmentSearch.BondsPerSPList[2*i];
4683 while (Binder->next != FragmentSearch.BondsPerSPList[2*i+1]) {
4684 Binder = Binder->next;
4685 SP ++;
4686 }
4687 }
4688 *out << Verbose(0) << "Total number of edges is " << SP << "." << endl;
4689 if (SP >= (Order-1)) {
4690 // start with root (push on fragment stack)
4691 *out << Verbose(0) << "Starting fragment generation with " << *FragmentSearch.Root << ", local nr is " << FragmentSearch.Root->nr << "." << endl;
4692 FragmentSearch.FragmentSet->clear();
4693 *out << Verbose(0) << "Preparing subset for this root and calling generator." << endl;
4694 // prepare the subset and call the generator
4695 BondsList = (bond **) Malloc(sizeof(bond *)*FragmentSearch.BondsPerSPCount[0], "molecule::PowerSetGenerator: **BondsList");
4696 BondsList[0] = FragmentSearch.BondsPerSPList[0]->next; // on SP level 0 there's only the root bond
4697
4698 SPFragmentGenerator(out, &FragmentSearch, 0, BondsList, FragmentSearch.BondsPerSPCount[0], Order);
4699
4700 Free((void **)&BondsList, "molecule::PowerSetGenerator: **BondsList");
4701 } else {
4702 *out << Verbose(0) << "Not enough total number of edges to build " << Order << "-body fragments." << endl;
4703 }
4704
4705 // as FragmentSearch structure is used only once, we don't have to clean it anymore
4706 // remove root from stack
4707 *out << Verbose(0) << "Removing root again from stack." << endl;
4708 FragmentSearch.FragmentSet->erase(FragmentSearch.Root->nr);
4709
4710 // free'ing the bonds lists
4711 *out << Verbose(0) << "Free'ing all found lists. and resetting index lists" << endl;
4712 for(int i=Order;i--;) {
4713 *out << Verbose(1) << "Current SP level is " << i << ": ";
4714 Binder = FragmentSearch.BondsPerSPList[2*i];
4715 while (Binder->next != FragmentSearch.BondsPerSPList[2*i+1]) {
4716 Binder = Binder->next;
4717 // *out << "Removing atom " << Binder->leftatom->nr << " and " << Binder->rightatom->nr << "." << endl; // make sure numbers are local
4718 FragmentSearch.ShortestPathList[Binder->leftatom->nr] = -1;
4719 FragmentSearch.ShortestPathList[Binder->rightatom->nr] = -1;
4720 }
4721 // delete added bonds
4722 cleanup(FragmentSearch.BondsPerSPList[2*i], FragmentSearch.BondsPerSPList[2*i+1]);
4723 // also start and end node
4724 *out << "cleaned." << endl;
4725 }
4726
4727 // return list
4728 *out << Verbose(0) << "End of PowerSetGenerator." << endl;
4729 return (FragmentSearch.FragmentCounter - Counter);
4730};
4731
4732/** Corrects the nuclei position if the fragment was created over the cell borders.
4733 * Scans all bonds, checks the distance, if greater than typical, we have a candidate for the correction.
4734 * We remove the bond whereafter the graph probably separates. Then, we translate the one component periodically
4735 * and re-add the bond. Looping on the distance check.
4736 * \param *out ofstream for debugging messages
4737 */
4738void molecule::ScanForPeriodicCorrection(ofstream *out)
4739{
4740 bond *Binder = NULL;
4741 bond *OtherBinder = NULL;
4742 atom *Walker = NULL;
4743 atom *OtherWalker = NULL;
4744 double *matrix = ReturnFullMatrixforSymmetric(cell_size);
4745 enum Shading *ColorList = NULL;
4746 double tmp;
4747 Vector Translationvector;
4748 //class StackClass<atom *> *CompStack = NULL;
4749 class StackClass<atom *> *AtomStack = new StackClass<atom *>(AtomCount);
4750 bool flag = true;
4751
4752 *out << Verbose(2) << "Begin of ScanForPeriodicCorrection." << endl;
4753
4754 ColorList = (enum Shading *) Malloc(sizeof(enum Shading)*AtomCount, "molecule::ScanForPeriodicCorrection: *ColorList");
4755 while (flag) {
4756 // remove bonds that are beyond bonddistance
4757 for(int i=NDIM;i--;)
4758 Translationvector.x[i] = 0.;
4759 // scan all bonds
4760 Binder = first;
4761 flag = false;
4762 while ((!flag) && (Binder->next != last)) {
4763 Binder = Binder->next;
4764 for (int i=NDIM;i--;) {
4765 tmp = fabs(Binder->leftatom->x.x[i] - Binder->rightatom->x.x[i]);
4766 //*out << Verbose(3) << "Checking " << i << "th distance of " << *Binder->leftatom << " to " << *Binder->rightatom << ": " << tmp << "." << endl;
4767 if (tmp > BondDistance) {
4768 OtherBinder = Binder->next; // note down binding partner for later re-insertion
4769 unlink(Binder); // unlink bond
4770 *out << Verbose(2) << "Correcting at bond " << *Binder << "." << endl;
4771 flag = true;
4772 break;
4773 }
4774 }
4775 }
4776 if (flag) {
4777 // create translation vector from their periodically modified distance
4778 for (int i=NDIM;i--;) {
4779 tmp = Binder->leftatom->x.x[i] - Binder->rightatom->x.x[i];
4780 if (fabs(tmp) > BondDistance)
4781 Translationvector.x[i] = (tmp < 0) ? +1. : -1.;
4782 }
4783 Translationvector.MatrixMultiplication(matrix);
4784 //*out << Verbose(3) << "Translation vector is ";
4785 Translationvector.Output(out);
4786 *out << endl;
4787 // apply to all atoms of first component via BFS
4788 for (int i=AtomCount;i--;)
4789 ColorList[i] = white;
4790 AtomStack->Push(Binder->leftatom);
4791 while (!AtomStack->IsEmpty()) {
4792 Walker = AtomStack->PopFirst();
4793 //*out << Verbose (3) << "Current Walker is: " << *Walker << "." << endl;
4794 ColorList[Walker->nr] = black; // mark as explored
4795 Walker->x.AddVector(&Translationvector); // translate
4796 for (int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++) { // go through all binding partners
4797 if (ListOfBondsPerAtom[Walker->nr][i] != Binder) {
4798 OtherWalker = ListOfBondsPerAtom[Walker->nr][i]->GetOtherAtom(Walker);
4799 if (ColorList[OtherWalker->nr] == white) {
4800 AtomStack->Push(OtherWalker); // push if yet unexplored
4801 }
4802 }
4803 }
4804 }
4805 // re-add bond
4806 link(Binder, OtherBinder);
4807 } else {
4808 *out << Verbose(3) << "No corrections for this fragment." << endl;
4809 }
4810 //delete(CompStack);
4811 }
4812
4813 // free allocated space from ReturnFullMatrixforSymmetric()
4814 delete(AtomStack);
4815 Free((void **)&ColorList, "molecule::ScanForPeriodicCorrection: *ColorList");
4816 Free((void **)&matrix, "molecule::ScanForPeriodicCorrection: *matrix");
4817 *out << Verbose(2) << "End of ScanForPeriodicCorrection." << endl;
4818};
4819
4820/** Blows the 6-dimensional \a cell_size array up to a full NDIM by NDIM matrix.
4821 * \param *symm 6-dim array of unique symmetric matrix components
4822 * \return allocated NDIM*NDIM array with the symmetric matrix
4823 */
4824double * molecule::ReturnFullMatrixforSymmetric(double *symm)
4825{
4826 double *matrix = (double *) Malloc(sizeof(double)*NDIM*NDIM, "molecule::ReturnFullMatrixforSymmetric: *matrix");
4827 matrix[0] = symm[0];
4828 matrix[1] = symm[1];
4829 matrix[2] = symm[3];
4830 matrix[3] = symm[1];
4831 matrix[4] = symm[2];
4832 matrix[5] = symm[4];
4833 matrix[6] = symm[3];
4834 matrix[7] = symm[4];
4835 matrix[8] = symm[5];
4836 return matrix;
4837};
4838
4839bool KeyCompare::operator() (const KeySet SubgraphA, const KeySet SubgraphB) const
4840{
4841 //cout << "my check is used." << endl;
4842 if (SubgraphA.size() < SubgraphB.size()) {
4843 return true;
4844 } else {
4845 if (SubgraphA.size() > SubgraphB.size()) {
4846 return false;
4847 } else {
4848 KeySet::iterator IteratorA = SubgraphA.begin();
4849 KeySet::iterator IteratorB = SubgraphB.begin();
4850 while ((IteratorA != SubgraphA.end()) && (IteratorB != SubgraphB.end())) {
4851 if ((*IteratorA) < (*IteratorB))
4852 return true;
4853 else if ((*IteratorA) > (*IteratorB)) {
4854 return false;
4855 } // else, go on to next index
4856 IteratorA++;
4857 IteratorB++;
4858 } // end of while loop
4859 }// end of check in case of equal sizes
4860 }
4861 return false; // if we reach this point, they are equal
4862};
4863
4864//bool operator < (KeySet SubgraphA, KeySet SubgraphB)
4865//{
4866// return KeyCompare(SubgraphA, SubgraphB);
4867//};
4868
4869/** Checking whether KeySet is not already present in Graph, if so just adds factor.
4870 * \param *out output stream for debugging
4871 * \param &set KeySet to insert
4872 * \param &graph Graph to insert into
4873 * \param *counter pointer to unique fragment count
4874 * \param factor energy factor for the fragment
4875 */
4876inline void InsertFragmentIntoGraph(ofstream *out, struct UniqueFragments *Fragment)
4877{
4878 GraphTestPair testGraphInsert;
4879
4880 testGraphInsert = Fragment->Leaflet->insert(GraphPair (*Fragment->FragmentSet,pair<int,double>(Fragment->FragmentCounter,Fragment->TEFactor))); // store fragment number and current factor
4881 if (testGraphInsert.second) {
4882 *out << Verbose(2) << "KeySet " << Fragment->FragmentCounter << " successfully inserted." << endl;
4883 Fragment->FragmentCounter++;
4884 } else {
4885 *out << Verbose(2) << "KeySet " << Fragment->FragmentCounter << " failed to insert, present fragment is " << ((*(testGraphInsert.first)).second).first << endl;
4886 ((*(testGraphInsert.first)).second).second += Fragment->TEFactor; // increase the "created" counter
4887 *out << Verbose(2) << "New factor is " << ((*(testGraphInsert.first)).second).second << "." << endl;
4888 }
4889};
4890//void inline InsertIntoGraph(ofstream *out, KeyStack &stack, Graph &graph, int *counter, double factor)
4891//{
4892// // copy stack contents to set and call overloaded function again
4893// KeySet set;
4894// for(KeyStack::iterator runner = stack.begin(); runner != stack.begin(); runner++)
4895// set.insert((*runner));
4896// InsertIntoGraph(out, set, graph, counter, factor);
4897//};
4898
4899/** Inserts each KeySet in \a graph2 into \a graph1.
4900 * \param *out output stream for debugging
4901 * \param graph1 first (dest) graph
4902 * \param graph2 second (source) graph
4903 * \param *counter keyset counter that gets increased
4904 */
4905inline void InsertGraphIntoGraph(ofstream *out, Graph &graph1, Graph &graph2, int *counter)
4906{
4907 GraphTestPair testGraphInsert;
4908
4909 for(Graph::iterator runner = graph2.begin(); runner != graph2.end(); runner++) {
4910 testGraphInsert = graph1.insert(GraphPair ((*runner).first,pair<int,double>((*counter)++,((*runner).second).second))); // store fragment number and current factor
4911 if (testGraphInsert.second) {
4912 *out << Verbose(2) << "KeySet " << (*counter)-1 << " successfully inserted." << endl;
4913 } else {
4914 *out << Verbose(2) << "KeySet " << (*counter)-1 << " failed to insert, present fragment is " << ((*(testGraphInsert.first)).second).first << endl;
4915 ((*(testGraphInsert.first)).second).second += (*runner).second.second;
4916 *out << Verbose(2) << "New factor is " << (*(testGraphInsert.first)).second.second << "." << endl;
4917 }
4918 }
4919};
4920
4921
4922/** Performs BOSSANOVA decomposition at selected sites, increasing the cutoff by one at these sites.
4923 * -# constructs a complete keyset of the molecule
4924 * -# In a loop over all possible roots from the given rootstack
4925 * -# increases order of root site
4926 * -# calls PowerSetGenerator with this order, the complete keyset and the rootkeynr
4927 * -# for all consecutive lower levels PowerSetGenerator is called with the suborder, the higher order keyset
4928as the restricted one and each site in the set as the root)
4929 * -# these are merged into a fragment list of keysets
4930 * -# All fragment lists (for all orders, i.e. from all destination fields) are merged into one list for return
4931 * Important only is that we create all fragments, it is not important if we create them more than once
4932 * as these copies are filtered out via use of the hash table (KeySet).
4933 * \param *out output stream for debugging
4934 * \param Fragment&*List list of already present keystacks (adaptive scheme) or empty list
4935 * \param &RootStack stack with all root candidates (unequal to each atom in complete molecule if adaptive scheme is applied)
4936 * \param *MinimumRingSize minimum ring size for each atom (molecule::Atomcount)
4937 * \return pointer to Graph list
4938 */
4939void molecule::FragmentBOSSANOVA(ofstream *out, Graph *&FragmentList, KeyStack &RootStack, int *MinimumRingSize)
4940{
4941 Graph ***FragmentLowerOrdersList = NULL;
4942 int NumLevels, NumMolecules, TotalNumMolecules = 0, *NumMoleculesOfOrder = NULL;
4943 int counter = 0, Order;
4944 int UpgradeCount = RootStack.size();
4945 KeyStack FragmentRootStack;
4946 int RootKeyNr, RootNr;
4947 struct UniqueFragments FragmentSearch;
4948
4949 *out << Verbose(0) << "Begin of FragmentBOSSANOVA." << endl;
4950
4951 // FragmentLowerOrdersList is a 2D-array of pointer to MoleculeListClass objects, one dimension represents the ANOVA expansion of a single order (i.e. 5)
4952 // with all needed lower orders that are subtracted, the other dimension is the BondOrder (i.e. from 1 to 5)
4953 NumMoleculesOfOrder = (int *) Malloc(sizeof(int)*UpgradeCount, "molecule::FragmentBOSSANOVA: *NumMoleculesOfOrder");
4954 FragmentLowerOrdersList = (Graph ***) Malloc(sizeof(Graph **)*UpgradeCount, "molecule::FragmentBOSSANOVA: ***FragmentLowerOrdersList");
4955
4956 // initialise the fragments structure
4957 FragmentSearch.ShortestPathList = (int *) Malloc(sizeof(int)*AtomCount, "molecule::PowerSetGenerator: *ShortestPathList");
4958 FragmentSearch.FragmentCounter = 0;
4959 FragmentSearch.FragmentSet = new KeySet;
4960 FragmentSearch.Root = FindAtom(RootKeyNr);
4961 for (int i=AtomCount;i--;) {
4962 FragmentSearch.ShortestPathList[i] = -1;
4963 }
4964
4965 // Construct the complete KeySet which we need for topmost level only (but for all Roots)
4966 atom *Walker = start;
4967 KeySet CompleteMolecule;
4968 while (Walker->next != end) {
4969 Walker = Walker->next;
4970 CompleteMolecule.insert(Walker->GetTrueFather()->nr);
4971 }
4972
4973 // this can easily be seen: if Order is 5, then the number of levels for each lower order is the total sum of the number of levels above, as
4974 // each has to be split up. E.g. for the second level we have one from 5th, one from 4th, two from 3th (which in turn is one from 5th, one from 4th),
4975 // hence we have overall four 2th order levels for splitting. This also allows for putting all into a single array (FragmentLowerOrdersList[])
4976 // with the order along the cells as this: 5433222211111111 for BondOrder 5 needing 16=pow(2,5-1) cells (only we use bit-shifting which is faster)
4977 RootNr = 0; // counts through the roots in RootStack
4978 while ((RootNr < UpgradeCount) && (!RootStack.empty())) {
4979 RootKeyNr = RootStack.front();
4980 RootStack.pop_front();
4981 Walker = FindAtom(RootKeyNr);
4982 // check cyclic lengths
4983 //if ((MinimumRingSize[Walker->GetTrueFather()->nr] != -1) && (Walker->GetTrueFather()->AdaptiveOrder+1 > MinimumRingSize[Walker->GetTrueFather()->nr])) {
4984 // *out << Verbose(0) << "Bond order " << Walker->GetTrueFather()->AdaptiveOrder << " of Root " << *Walker << " greater than or equal to Minimum Ring size of " << MinimumRingSize << " found is not allowed." << endl;
4985 //} else
4986 {
4987 // increase adaptive order by one
4988 Walker->GetTrueFather()->AdaptiveOrder++;
4989 Order = Walker->AdaptiveOrder = Walker->GetTrueFather()->AdaptiveOrder;
4990
4991 // initialise Order-dependent entries of UniqueFragments structure
4992 FragmentSearch.BondsPerSPList = (bond **) Malloc(sizeof(bond *)*Order*2, "molecule::PowerSetGenerator: ***BondsPerSPList");
4993 FragmentSearch.BondsPerSPCount = (int *) Malloc(sizeof(int)*Order, "molecule::PowerSetGenerator: *BondsPerSPCount");
4994 for (int i=Order;i--;) {
4995 FragmentSearch.BondsPerSPList[2*i] = new bond(); // start node
4996 FragmentSearch.BondsPerSPList[2*i+1] = new bond(); // end node
4997 FragmentSearch.BondsPerSPList[2*i]->next = FragmentSearch.BondsPerSPList[2*i+1]; // intertwine these two
4998 FragmentSearch.BondsPerSPList[2*i+1]->previous = FragmentSearch.BondsPerSPList[2*i];
4999 FragmentSearch.BondsPerSPCount[i] = 0;
5000 }
5001
5002 // allocate memory for all lower level orders in this 1D-array of ptrs
5003 NumLevels = 1 << (Order-1); // (int)pow(2,Order);
5004 FragmentLowerOrdersList[RootNr] = (Graph **) Malloc(sizeof(Graph *)*NumLevels, "molecule::FragmentBOSSANOVA: **FragmentLowerOrdersList[]");
5005 for (int i=0;i<NumLevels;i++)
5006 FragmentLowerOrdersList[RootNr][i] = NULL;
5007
5008 // create top order where nothing is reduced
5009 *out << Verbose(0) << "==============================================================================================================" << endl;
5010 *out << Verbose(0) << "Creating KeySets of Bond Order " << Order << " for " << *Walker << ", " << (RootStack.size()-RootNr) << " Roots remaining." << endl; // , NumLevels is " << NumLevels << "
5011
5012 // Create list of Graphs of current Bond Order (i.e. F_{ij})
5013 FragmentLowerOrdersList[RootNr][0] = new Graph;
5014 FragmentSearch.TEFactor = 1.;
5015 FragmentSearch.Leaflet = FragmentLowerOrdersList[RootNr][0]; // set to insertion graph
5016 FragmentSearch.Root = Walker;
5017 NumMoleculesOfOrder[RootNr] = PowerSetGenerator(out, Walker->AdaptiveOrder, FragmentSearch, CompleteMolecule);
5018 *out << Verbose(1) << "Number of resulting KeySets is: " << NumMoleculesOfOrder[RootNr] << "." << endl;
5019 if (NumMoleculesOfOrder[RootNr] != 0) {
5020 NumMolecules = 0;
5021
5022 // we don't have to dive into suborders! These keysets are all already created on lower orders!
5023 // this was all ancient stuff, when we still depended on the TEFactors (and for those the suborders were needed)
5024
5025// if ((NumLevels >> 1) > 0) {
5026// // create lower order fragments
5027// *out << Verbose(0) << "Creating list of unique fragments of lower Bond Order terms to be subtracted." << endl;
5028// Order = Walker->AdaptiveOrder;
5029// for (int source=0;source<(NumLevels >> 1);source++) { // 1-terms don't need any more splitting, that's why only half is gone through (shift again)
5030// // step down to next order at (virtual) boundary of powers of 2 in array
5031// while (source >= (1 << (Walker->AdaptiveOrder-Order))) // (int)pow(2,Walker->AdaptiveOrder-Order))
5032// Order--;
5033// *out << Verbose(0) << "Current Order is: " << Order << "." << endl;
5034// for (int SubOrder=Order-1;SubOrder>0;SubOrder--) {
5035// int dest = source + (1 << (Walker->AdaptiveOrder-(SubOrder+1)));
5036// *out << Verbose(0) << "--------------------------------------------------------------------------------------------------------------" << endl;
5037// *out << Verbose(0) << "Current SubOrder is: " << SubOrder << " with source " << source << " to destination " << dest << "." << endl;
5038//
5039// // every molecule is split into a list of again (Order - 1) molecules, while counting all molecules
5040// //*out << Verbose(1) << "Splitting the " << (*FragmentLowerOrdersList[RootNr][source]).size() << " molecules of the " << source << "th cell in the array." << endl;
5041// //NumMolecules = 0;
5042// FragmentLowerOrdersList[RootNr][dest] = new Graph;
5043// for(Graph::iterator runner = (*FragmentLowerOrdersList[RootNr][source]).begin();runner != (*FragmentLowerOrdersList[RootNr][source]).end(); runner++) {
5044// for (KeySet::iterator sprinter = (*runner).first.begin();sprinter != (*runner).first.end(); sprinter++) {
5045// Graph TempFragmentList;
5046// FragmentSearch.TEFactor = -(*runner).second.second;
5047// FragmentSearch.Leaflet = &TempFragmentList; // set to insertion graph
5048// FragmentSearch.Root = FindAtom(*sprinter);
5049// NumMoleculesOfOrder[RootNr] += PowerSetGenerator(out, SubOrder, FragmentSearch, (*runner).first);
5050// // insert new keysets FragmentList into FragmentLowerOrdersList[Walker->AdaptiveOrder-1][dest]
5051// *out << Verbose(1) << "Merging resulting key sets with those present in destination " << dest << "." << endl;
5052// InsertGraphIntoGraph(out, *FragmentLowerOrdersList[RootNr][dest], TempFragmentList, &NumMolecules);
5053// }
5054// }
5055// *out << Verbose(1) << "Number of resulting molecules for SubOrder " << SubOrder << " is: " << NumMolecules << "." << endl;
5056// }
5057// }
5058// }
5059 } else {
5060 Walker->GetTrueFather()->MaxOrder = true;
5061// *out << Verbose(1) << "Hence, we don't dive into SubOrders ... " << endl;
5062 }
5063 // now, we have completely filled each cell of FragmentLowerOrdersList[] for the current Walker->AdaptiveOrder
5064 //NumMoleculesOfOrder[Walker->AdaptiveOrder-1] = NumMolecules;
5065 TotalNumMolecules += NumMoleculesOfOrder[RootNr];
5066// *out << Verbose(1) << "Number of resulting molecules for Order " << (int)Walker->GetTrueFather()->AdaptiveOrder << " is: " << NumMoleculesOfOrder[RootNr] << "." << endl;
5067 RootStack.push_back(RootKeyNr); // put back on stack
5068 RootNr++;
5069
5070 // free Order-dependent entries of UniqueFragments structure for next loop cycle
5071 Free((void **)&FragmentSearch.BondsPerSPCount, "molecule::PowerSetGenerator: *BondsPerSPCount");
5072 for (int i=Order;i--;) {
5073 delete(FragmentSearch.BondsPerSPList[2*i]);
5074 delete(FragmentSearch.BondsPerSPList[2*i+1]);
5075 }
5076 Free((void **)&FragmentSearch.BondsPerSPList, "molecule::PowerSetGenerator: ***BondsPerSPList");
5077 }
5078 }
5079 *out << Verbose(0) << "==============================================================================================================" << endl;
5080 *out << Verbose(1) << "Total number of resulting molecules is: " << TotalNumMolecules << "." << endl;
5081 *out << Verbose(0) << "==============================================================================================================" << endl;
5082
5083 // cleanup FragmentSearch structure
5084 Free((void **)&FragmentSearch.ShortestPathList, "molecule::PowerSetGenerator: *ShortestPathList");
5085 delete(FragmentSearch.FragmentSet);
5086
5087 // now, FragmentLowerOrdersList is complete, it looks - for BondOrder 5 - as this (number is the ANOVA Order of the terms therein)
5088 // 5433222211111111
5089 // 43221111
5090 // 3211
5091 // 21
5092 // 1
5093
5094 // Subsequently, we combine all into a single list (FragmentList)
5095
5096 *out << Verbose(0) << "Combining the lists of all orders per order and finally into a single one." << endl;
5097 if (FragmentList == NULL) {
5098 FragmentList = new Graph;
5099 counter = 0;
5100 } else {
5101 counter = FragmentList->size();
5102 }
5103 RootNr = 0;
5104 while (!RootStack.empty()) {
5105 RootKeyNr = RootStack.front();
5106 RootStack.pop_front();
5107 Walker = FindAtom(RootKeyNr);
5108 NumLevels = 1 << (Walker->AdaptiveOrder - 1);
5109 for(int i=0;i<NumLevels;i++) {
5110 if (FragmentLowerOrdersList[RootNr][i] != NULL) {
5111 InsertGraphIntoGraph(out, *FragmentList, (*FragmentLowerOrdersList[RootNr][i]), &counter);
5112 delete(FragmentLowerOrdersList[RootNr][i]);
5113 }
5114 }
5115 Free((void **)&FragmentLowerOrdersList[RootNr], "molecule::FragmentBOSSANOVA: **FragmentLowerOrdersList[]");
5116 RootNr++;
5117 }
5118 Free((void **)&FragmentLowerOrdersList, "molecule::FragmentBOSSANOVA: ***FragmentLowerOrdersList");
5119 Free((void **)&NumMoleculesOfOrder, "molecule::FragmentBOSSANOVA: *NumMoleculesOfOrder");
5120
5121 *out << Verbose(0) << "End of FragmentBOSSANOVA." << endl;
5122};
5123
5124/** Comparison function for GSL heapsort on distances in two molecules.
5125 * \param *a
5126 * \param *b
5127 * \return <0, \a *a less than \a *b, ==0 if equal, >0 \a *a greater than \a *b
5128 */
5129inline int CompareDoubles (const void * a, const void * b)
5130{
5131 if (*(double *)a > *(double *)b)
5132 return -1;
5133 else if (*(double *)a < *(double *)b)
5134 return 1;
5135 else
5136 return 0;
5137};
5138
5139/** Determines whether two molecules actually contain the same atoms and coordination.
5140 * \param *out output stream for debugging
5141 * \param *OtherMolecule the molecule to compare this one to
5142 * \param threshold upper limit of difference when comparing the coordination.
5143 * \return NULL - not equal, otherwise an allocated (molecule::AtomCount) permutation map of the atom numbers (which corresponds to which)
5144 */
5145int * molecule::IsEqualToWithinThreshold(ofstream *out, molecule *OtherMolecule, double threshold)
5146{
5147 int flag;
5148 double *Distances = NULL, *OtherDistances = NULL;
5149 Vector CenterOfGravity, OtherCenterOfGravity;
5150 size_t *PermMap = NULL, *OtherPermMap = NULL;
5151 int *PermutationMap = NULL;
5152 atom *Walker = NULL;
5153 bool result = true; // status of comparison
5154
5155 *out << Verbose(3) << "Begin of IsEqualToWithinThreshold." << endl;
5156 /// first count both their atoms and elements and update lists thereby ...
5157 //*out << Verbose(0) << "Counting atoms, updating list" << endl;
5158 CountAtoms(out);
5159 OtherMolecule->CountAtoms(out);
5160 CountElements();
5161 OtherMolecule->CountElements();
5162
5163 /// ... and compare:
5164 /// -# AtomCount
5165 if (result) {
5166 if (AtomCount != OtherMolecule->AtomCount) {
5167 *out << Verbose(4) << "AtomCounts don't match: " << AtomCount << " == " << OtherMolecule->AtomCount << endl;
5168 result = false;
5169 } else *out << Verbose(4) << "AtomCounts match: " << AtomCount << " == " << OtherMolecule->AtomCount << endl;
5170 }
5171 /// -# ElementCount
5172 if (result) {
5173 if (ElementCount != OtherMolecule->ElementCount) {
5174 *out << Verbose(4) << "ElementCount don't match: " << ElementCount << " == " << OtherMolecule->ElementCount << endl;
5175 result = false;
5176 } else *out << Verbose(4) << "ElementCount match: " << ElementCount << " == " << OtherMolecule->ElementCount << endl;
5177 }
5178 /// -# ElementsInMolecule
5179 if (result) {
5180 for (flag=MAX_ELEMENTS;flag--;) {
5181 //*out << Verbose(5) << "Element " << flag << ": " << ElementsInMolecule[flag] << " <-> " << OtherMolecule->ElementsInMolecule[flag] << "." << endl;
5182 if (ElementsInMolecule[flag] != OtherMolecule->ElementsInMolecule[flag])
5183 break;
5184 }
5185 if (flag < MAX_ELEMENTS) {
5186 *out << Verbose(4) << "ElementsInMolecule don't match." << endl;
5187 result = false;
5188 } else *out << Verbose(4) << "ElementsInMolecule match." << endl;
5189 }
5190 /// then determine and compare center of gravity for each molecule ...
5191 if (result) {
5192 *out << Verbose(5) << "Calculating Centers of Gravity" << endl;
5193 DetermineCenter(CenterOfGravity);
5194 OtherMolecule->DetermineCenter(OtherCenterOfGravity);
5195 *out << Verbose(5) << "Center of Gravity: ";
5196 CenterOfGravity.Output(out);
5197 *out << endl << Verbose(5) << "Other Center of Gravity: ";
5198 OtherCenterOfGravity.Output(out);
5199 *out << endl;
5200 if (CenterOfGravity.Distance(&OtherCenterOfGravity) > threshold) {
5201 *out << Verbose(4) << "Centers of gravity don't match." << endl;
5202 result = false;
5203 }
5204 }
5205
5206 /// ... then make a list with the euclidian distance to this center for each atom of both molecules
5207 if (result) {
5208 *out << Verbose(5) << "Calculating distances" << endl;
5209 Distances = (double *) Malloc(sizeof(double)*AtomCount, "molecule::IsEqualToWithinThreshold: Distances");
5210 OtherDistances = (double *) Malloc(sizeof(double)*AtomCount, "molecule::IsEqualToWithinThreshold: OtherDistances");
5211 Walker = start;
5212 while (Walker->next != end) {
5213 Walker = Walker->next;
5214 Distances[Walker->nr] = CenterOfGravity.Distance(&Walker->x);
5215 }
5216 Walker = OtherMolecule->start;
5217 while (Walker->next != OtherMolecule->end) {
5218 Walker = Walker->next;
5219 OtherDistances[Walker->nr] = OtherCenterOfGravity.Distance(&Walker->x);
5220 }
5221
5222 /// ... sort each list (using heapsort (o(N log N)) from GSL)
5223 *out << Verbose(5) << "Sorting distances" << endl;
5224 PermMap = (size_t *) Malloc(sizeof(size_t)*AtomCount, "molecule::IsEqualToWithinThreshold: *PermMap");
5225 OtherPermMap = (size_t *) Malloc(sizeof(size_t)*AtomCount, "molecule::IsEqualToWithinThreshold: *OtherPermMap");
5226 gsl_heapsort_index (PermMap, Distances, AtomCount, sizeof(double), CompareDoubles);
5227 gsl_heapsort_index (OtherPermMap, OtherDistances, AtomCount, sizeof(double), CompareDoubles);
5228 PermutationMap = (int *) Malloc(sizeof(int)*AtomCount, "molecule::IsEqualToWithinThreshold: *PermutationMap");
5229 *out << Verbose(5) << "Combining Permutation Maps" << endl;
5230 for(int i=AtomCount;i--;)
5231 PermutationMap[PermMap[i]] = (int) OtherPermMap[i];
5232
5233 /// ... and compare them step by step, whether the difference is individiually(!) below \a threshold for all
5234 *out << Verbose(4) << "Comparing distances" << endl;
5235 flag = 0;
5236 for (int i=0;i<AtomCount;i++) {
5237 *out << Verbose(5) << "Distances: |" << Distances[PermMap[i]] << " - " << OtherDistances[OtherPermMap[i]] << "| = " << fabs(Distances[PermMap[i]] - OtherDistances[OtherPermMap[i]]) << " ?<? " << threshold << endl;
5238 if (fabs(Distances[PermMap[i]] - OtherDistances[OtherPermMap[i]]) > threshold)
5239 flag = 1;
5240 }
5241 Free((void **)&PermMap, "molecule::IsEqualToWithinThreshold: *PermMap");
5242 Free((void **)&OtherPermMap, "molecule::IsEqualToWithinThreshold: *OtherPermMap");
5243
5244 /// free memory
5245 Free((void **)&Distances, "molecule::IsEqualToWithinThreshold: Distances");
5246 Free((void **)&OtherDistances, "molecule::IsEqualToWithinThreshold: OtherDistances");
5247 if (flag) { // if not equal
5248 Free((void **)&PermutationMap, "molecule::IsEqualToWithinThreshold: *PermutationMap");
5249 result = false;
5250 }
5251 }
5252 /// return pointer to map if all distances were below \a threshold
5253 *out << Verbose(3) << "End of IsEqualToWithinThreshold." << endl;
5254 if (result) {
5255 *out << Verbose(3) << "Result: Equal." << endl;
5256 return PermutationMap;
5257 } else {
5258 *out << Verbose(3) << "Result: Not equal." << endl;
5259 return NULL;
5260 }
5261};
5262
5263/** Returns an index map for two father-son-molecules.
5264 * The map tells which atom in this molecule corresponds to which one in the other molecul with their fathers.
5265 * \param *out output stream for debugging
5266 * \param *OtherMolecule corresponding molecule with fathers
5267 * \return allocated map of size molecule::AtomCount with map
5268 * \todo make this with a good sort O(n), not O(n^2)
5269 */
5270int * molecule::GetFatherSonAtomicMap(ofstream *out, molecule *OtherMolecule)
5271{
5272 atom *Walker = NULL, *OtherWalker = NULL;
5273 *out << Verbose(3) << "Begin of GetFatherAtomicMap." << endl;
5274 int *AtomicMap = (int *) Malloc(sizeof(int)*AtomCount, "molecule::GetAtomicMap: *AtomicMap"); //Calloc
5275 for (int i=AtomCount;i--;)
5276 AtomicMap[i] = -1;
5277 if (OtherMolecule == this) { // same molecule
5278 for (int i=AtomCount;i--;) // no need as -1 means already that there is trivial correspondence
5279 AtomicMap[i] = i;
5280 *out << Verbose(4) << "Map is trivial." << endl;
5281 } else {
5282 *out << Verbose(4) << "Map is ";
5283 Walker = start;
5284 while (Walker->next != end) {
5285 Walker = Walker->next;
5286 if (Walker->father == NULL) {
5287 AtomicMap[Walker->nr] = -2;
5288 } else {
5289 OtherWalker = OtherMolecule->start;
5290 while (OtherWalker->next != OtherMolecule->end) {
5291 OtherWalker = OtherWalker->next;
5292 //for (int i=0;i<AtomCount;i++) { // search atom
5293 //for (int j=0;j<OtherMolecule->AtomCount;j++) {
5294 //*out << Verbose(4) << "Comparing father " << Walker->father << " with the other one " << OtherWalker->father << "." << endl;
5295 if (Walker->father == OtherWalker)
5296 AtomicMap[Walker->nr] = OtherWalker->nr;
5297 }
5298 }
5299 *out << AtomicMap[Walker->nr] << "\t";
5300 }
5301 *out << endl;
5302 }
5303 *out << Verbose(3) << "End of GetFatherAtomicMap." << endl;
5304 return AtomicMap;
5305};
5306
5307/** Stores the temperature evaluated from velocities in molecule::Trajectories.
5308 * We simply use the formula equivaleting temperature and kinetic energy:
5309 * \f$k_B T = \sum_i m_i v_i^2\f$
5310 * \param *out output stream for debugging
5311 * \param startstep first MD step in molecule::Trajectories
5312 * \param endstep last plus one MD step in molecule::Trajectories
5313 * \param *output output stream of temperature file
5314 * \return file written (true), failure on writing file (false)
5315 */
5316bool molecule::OutputTemperatureFromTrajectories(ofstream *out, int startstep, int endstep, ofstream *output)
5317{
5318 double temperature;
5319 atom *Walker = NULL;
5320 // test stream
5321 if (output == NULL)
5322 return false;
5323 else
5324 *output << "# Step Temperature [K] Temperature [a.u.]" << endl;
5325 for (int step=startstep;step < endstep; step++) { // loop over all time steps
5326 temperature = 0.;
5327 Walker = start;
5328 while (Walker->next != end) {
5329 Walker = Walker->next;
5330 for (int i=NDIM;i--;)
5331 temperature += Walker->type->mass * Trajectories[Walker].U.at(step).x[i]* Trajectories[Walker].U.at(step).x[i];
5332 }
5333 *output << step << "\t" << temperature*AtomicEnergyToKelvin << "\t" << temperature << endl;
5334 }
5335 return true;
5336};
Note: See TracBrowser for help on using the repository browser.