Changeset 274d45 for tests/regression/Filling/1/post/test.conf

Timestamp:

Jun 19, 2010, 3:54:47 PM (15 years ago)

Author:

Frederik Heber <heber@…>

Branches:

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

Children:

bf0edf

Parents:

ad2b411

git-author:

Frederik Heber <heber@…> (06/19/10 12:35:01)

git-committer:

Frederik Heber <heber@…> (06/19/10 15:54:47)

Message:

FIX: Atoms were stored not in the sequence they were loaded.

1. The main problem is molecule::atomSet which is a set<atom *>, i.e. atoms are sorted by their appearance in memory. As memory need not be allocated sequentially, this gives rise to extreme arbitririty which is not desired. Instead the atoms should be stored in the sequence they were loaded/created. The solution is as follows:

• config::SaveAll()
  • uses SimpleMultiMerge() instead of SimpleMultiAdd(). I.e. no atoms are copied (with new ids) but are relinked into a single molecule.
  • also, in case of only a single molecule in the list, we don't merge anything to avoid mol id changes.
  • MaxMinSteps were set according to MaxPsiDouble overriding correct results from molecule::CalculateOrbitals()
• molecule::atomSet is now a list<atom *>
• molecule::atomIds is a new set<atomId_t> (atomIdSet) which controls that (global) ids remain unique in the no more Atomset's set (but list)
• molecule::erase() erases also in atomIds
• molecule::insert() checks whether id is present by atomIds
• molecule::find() as std::list does not have a find, we just go through the list until the object is found (or not), this may be speeded up by another internal list.
• molecule::InternalPointer made lots of confusion as virtual function GoToFirst() is const, hence begin() (needed therein) returns const_iterator, which then cannot be simply re-cast into an iterator: We make it a pointer, reinterpret_cast the pointer and reference it back. Although InternalPointer is mutable, the compiler cannot use the non-const function begin() (it cannot be made const, as overloading is not allowed). (this is noted in the code also extensively.)
• molecule::containsAtom() does not use count but the new find, as it returns only boolean anyway.
• rewrote MoleculeListClass::SimpleMerge() to get rid of the extra iterator, as we remove all atoms in the end anyway.
• FIX: MoleculeListClass::SimpleMultiMerge() - the created mol was not inserted into the moleculelist in the end, although it is the only one to remain.

2. All other databases had missing headers with respect to those stored in elements_db.cpp. Hence, valence of hydrogen was not parsed and this caused several failures in CalculateOrbitals() (Psi numbers and MaxMinSetp in pcp conf file).

3. Subsequenytly, various test cases (12, 21, 30, 31, 36-38, 39) were broken. This had two reasons:

• Seemingly, CalculateOrbitals() was broken before hence MaxMinStep (pcp config) and MaxPsiDouble/PsiMaxNn[Up|Down] were always 0. (10-21,30-31,39)
• As the order is now correct, fixes from commits c9217161ec2a5d5db508557fe98a32068461f45b and 22a6da8380911571debebd69444d2615450bbbd8 were obselete and have been reverted (order of the Ion?_Type...): Molecules/6 (30), Molecules/7 (31), Filling/1 (39)
• Due to different ordering, Tesselation/3 (38) had completely different .dat file (though same tesselation)
• r3d had small differences, mostly order or epsilon (0 not being 0 but ..-e16), hence diffing was deactivated, as r3d is deprecated anyway (since vmd can render triangles as well and better).

Signed-off-by: Frederik Heber <heber@…>

File:

• tests/regression/Filling/1/post/test.conf (modified) (3 diffs)

tests/regression/Filling/1/post/test.conf

@@ -35 +35 @@
 RelEpsTotalE    1e-07   # relative change in total energy
 RelEpsKineticE  1e-05   # relative change in kinetic energy
-MaxMinStopStep  0       # check every ..th steps
+MaxMinStopStep  1350    # check every ..th steps
 MaxMinGapStopStep       1       # check every ..th steps
 
@@ -42 +42 @@
 InitRelEpsTotalE        1e-05   # relative change in total energy
 InitRelEpsKineticE      0.0001  # relative change in kinetic energy
-InitMaxMinStopStep      0       # check every ..th steps
+InitMaxMinStopStep      1350    # check every ..th steps
 InitMaxMinGapStopStep   1       # check every ..th steps
 
@@ -55 +55 @@
 RiemannTensor   0       # (Use metric)
 PsiType         0       # 0 - doubly occupied, 1 - SpinUp,SpinDown
-MaxPsiDouble    0       # here: specifying both maximum number of SpinUp- and -Down-states
-PsiMaxNoUp      0       # here: specifying maximum number of SpinUp-states
-PsiMaxNoDown    0       # here: specifying maximum number of SpinDown-states
+MaxPsiDouble    1350    # here: specifying both maximum number of SpinUp- and -Down-states
+PsiMaxNoUp      1350    # here: specifying maximum number of SpinUp-states
+PsiMaxNoDown    1350    # here: specifying maximum number of SpinDown-states
 AddPsis         0       # Additional unoccupied Psis for bandgap determination