source: src/FunctionApproximation/Extractors.hpp@ ffa69c

/*
 * Extractors.hpp
 *
 *  Created on: 15.10.2012
 *      Author: heber
 */

#ifndef TRAININGDATA_EXTRACTORS_HPP_
#define TRAININGDATA_EXTRACTORS_HPP_

// include config.h
#ifdef HAVE_CONFIG_H
#include <config.h>
#endif

#include <boost/function.hpp>

#include "Fragmentation/Summation/SetValues/Fragment.hpp"
#include "FunctionApproximation/FunctionModel.hpp"

class Fragment;

/** Namespace containing all simple extractor functions.
 *
 * Extractor functions extract distances from a given fragment matching with
 * a given set of particle types (i.e. elements, e.h. H2O).
 * Filter functions extract a subset of distances from a given set of distances
 * to be used with a specific model.
 *
 * To this end, each FunctionModel has both a filter and an extractor function.
 *
 * The functions in this namespace act as helpers or basic building blocks in
 * constructing such filters and extractors.
 *
 */
namespace Extractors {
  typedef Fragment::charges_t::const_iterator chargeiter_t;
  typedef std::vector<chargeiter_t> chargeiters_t;

  typedef size_t count_t;
  typedef Fragment::charge_t element_t;
  typedef std::map< element_t, count_t> elementcounts_t;
  typedef std::map< element_t, chargeiters_t > elementtargets_t;
  typedef std::vector< chargeiters_t > targets_per_combination_t;
  //!> typedef for particle designation
  typedef int ParticleType_t;
  //!> typedef for a vector of particle designations
  typedef std::vector<ParticleType_t> ParticleTypes_t;

  /** Namespace for some internal helper functions.
   *
   */
  namespace _detail {
    /** Gather all distance arguments from the same aligned vector of charges.
     *
     * Basically, we filter the positions indicated by the targets but
     * from a different vector that has the same layout.
     *
     * \param positions all nuclei positions
     * \param charges all nuclei charges
     * \param targets iterators on charges
     * \return filtered distance arguments
     */
    FunctionModel::arguments_t gatherDistancesFromTargets(
        const Fragment::positions_t& positions,
        const Fragment::charges_t& charges,
        const chargeiters_t &targets,
        const size_t globalid
        );

    /** Gather all positions from the same aligned vector of charges.
     *
     * Basically, we filter the positions indicated by the targets but
     * from a different vector that has the same layout.
     *
     * \param positions all nuclei positions
     * \param charges all nuclei charges
     * \param targets iterators on charges
     * \return filtered positions
     */
    Fragment::positions_t gatherPositionsFromTargets(
        const Fragment::positions_t& positions,
        const Fragment::charges_t& charges,
        const chargeiters_t& targets
        );

    /** Counts all same elements in the vector and places into map of elements.
     *
     * \param elements vector of elements
     * \return count of same element in vector
     */
    elementcounts_t getElementCounts(
        const Fragment::charges_t elements
        );

    /** Gather iterators to the elements related to the desired elementcounts.
     *
     * \param charges charges wherein to search for the elements
     * \param elementcounts number of desired hits per element
     * \return iterators equal to the initial vector of elements
     */
    elementtargets_t convertElementcountsToTargets(
        const Fragment::charges_t &charges,
        const elementcounts_t &elementcounts
        );

    /** Convert the alignment back to as it was in the original vector.
     *
     * We lost the information by storing it in a map. Hence, we need this
     * final step.
     *
     * \param elementtargets targets as they are in the map \a elementcounts
     * \param elements the original order of the elements
     * \param elementcounts the count per element for debugging checks
     * \return vector of targets in the order as they are in \a element
     */
    chargeiters_t realignElementtargets(
        const elementtargets_t &elementtargets,
        const Fragment::charges_t elements,
        const elementcounts_t &elementcounts
        );

    /** Searches for desired elements in charges in a unique manner.
     *
     * The idea is to have e.g. a fragment with charges 8,1,1,2 and
     * elements as 1,8,1 (e.g. for an angle HOH) and we get the
     * chargeiters in the desired manner on indices: 1,0,3.
     *
     * \param charges charges to look through
     * \param elements vector of elements to find
     */
    chargeiters_t
    gatherTargetsFromFragment(
        const Fragment::charges_t& charges,
        const Fragment::charges_t elements
        );

    /** Brings all charges together in a map.
     *
     * @param charges charges as possible keys and their iterators as values in the map
     * @param elements list of charges to pick as keys
     * @return map of key and a vector of charge iterators
     */
    elementtargets_t convertChargesToTargetMap(
        const Fragment::charges_t& charges,
        Fragment::charges_t elements
        );

    /** Brings combinatorially together desired list of \a charges and \a targets.
     *
     * @param charges list of desired charges
     * @param elementtargets map of available targets per element
     * @return vector of chargeiters with all unique combinations
     */
    targets_per_combination_t
    CombineChargesAndTargets(
        const Fragment::charges_t& charges,
        const elementtargets_t& elementtargets
        );

    /** Recursive function to pick the next target.
     *
     * This is used by \sa CombineChargesAndTargets()
     *
     * @param charges set of charges, reduced by one per recursion
     * @param elementtargets targets, map of targets to pick from
     * @param currenttargets current set of targets, "global" through recursion
     * @param addFunction bound function to add a set when complete
     */
    void pickLastElementAsTarget(
        Fragment::charges_t elements,
        elementtargets_t elementtargets,
        chargeiters_t& currenttargets,
        boost::function<void (const chargeiters_t &currenttargets)> &addFunction
        );

    /** Converts a list of chargeiters to a list of respective arguments.
     *
     * @param positions positions from fragment
     * @param charges charges associated to each element in \a positions
     * @param combinations vector of chargeiters
     * \param globalid refers to the index within the global set of configurations
     * @return list of arguments
     */
    FunctionModel::arguments_t
    convertTargetsToArguments(
        const Fragment::positions_t& positions,
        const Fragment::charges_t& charges,
        const targets_per_combination_t combinations,
        const size_t globalid
        );
  }

  /** Gather all distances from a given set of positions.
   *
   * \param positions all nuclei positions
   * \param charges all nuclei charges
   * \param globalid index to associated in argument_t with
   * \return vector of argument_ , each with a distance
   */
  FunctionModel::arguments_t
  gatherAllDistanceArguments(
      const Fragment::positions_t& positions,
      const Fragment::charges_t& charges,
      const size_t globalid);

  /** Gather all distances from a given set of positions.
   *
   *  Here, we only return one of the two equal distances.
   *
   * \param positions all nuclei positions
   * \param charges all nuclei charges
   * \param globalid index to associated in argument_t with
   * \return vector of argument_ , each with a distance
   */
  FunctionModel::arguments_t
  gatherAllSymmetricDistanceArguments(
      const Fragment::positions_t& positions,
      const Fragment::charges_t& charges,
      const size_t globalid);

  /** Simple extractor of all unique pair distances of a given \a fragment.
   *
   * \param positions all nuclei positions
   * \param charges all nuclei charges
   * \param index index refers to the index within the global set of configurations
   * \return vector of of argument_t containing all found distances
   */
  inline FunctionModel::arguments_t gatherAllDistances(
      const Fragment::positions_t& positions,
      const Fragment::charges_t& charges,
      const size_t index
      ) {
    // get distance out of Fragment
    return gatherAllDistanceArguments(positions, charges, index);
  }

  /** Simple extractor of all unique pair distances of a given \a fragment, where
   * the first index is less than the second one.
   *
   * \param positions all nuclei positions
   * \param charges all nuclei charges
   * \param index index refers to the index within the global set of configurations
   * \return vector of of argument_t containing all found distances
   */
  inline FunctionModel::arguments_t gatherAllSymmetricDistances(
      const Fragment::positions_t& positions,
      const Fragment::charges_t& charges,
      const size_t index
      ) {
    // get distance out of Fragment
    return gatherAllSymmetricDistanceArguments(positions, charges, index);
  }

  /** Filters only those positions out of given \a fragment that match \a elements.
   *
   * \param positions all nuclei positions
   * \param charges all nuclei charges
   * \param elements tuple of desired elements
   * \return vector of positions_t containing
   */
  Fragment::positions_t gatherPositionsFromFragment(
      const Fragment::positions_t positions,
      const Fragment::charges_t charges,
      const Fragment::charges_t& elements
      );

  /** Filters only those distances out of given \a fragment that match \a elements.
   *
   * \param positions all nuclei positions
   * \param charges all nuclei charges
   * \param elements tuple of desired elements
   * \param globalid refers to the index within the global set of configurations
   * \return vector of arguments_t containing those matched with elements
   */
  FunctionModel::arguments_t gatherDistancesFromFragment(
      const Fragment::positions_t positions,
      const Fragment::charges_t charges,
      const Fragment::charges_t& elements,
      const size_t globalid
      );

  /** Gather all combinations of charges as distance arguments from the fragment.
   *
   * E.g. we have a water fragment, i.e. (8,1,1) and we we want elements (8,1),
   * then two arguments are returned, first to second and first to third.
   *
   * With \sa gatherDistancesFromFragment() only the first distance would be
   * returned.
   *
   * @param positions positions in fragment
   * @param charges charges in fragment
   * @param elements list of desired elements
   * @param globalid some global id to discern training data tuples
   * @return list of arguments with distances
   */
  FunctionModel::arguments_t gatherAllDistancesFromFragment(
      const Fragment::positions_t& positions,
      const Fragment::charges_t& charges,
      const Fragment::charges_t elements,
      const size_t globalid
      );

  /** Reorder arguments by increasing distance.
   *
   * \param listargs list of arguments to reorder each
   * \return reordered args
   */
  FunctionModel::list_of_arguments_t reorderArgumentsByIncreasingDistance(
      const FunctionModel::list_of_arguments_t &listargs
      );

  /** Reorder the arguments to bring adjacent ones together.
   *
   * After filtering via particle types arguments related via same indices
   * must not necessarily be contained in the same bunch. This reordering
   * is done here, preserving the alignment given in
   * \sa filterArgumentsByParticleTypes()
   *
   * \param listargs list of arguments to reorder each
   * \param _types particle type vector
   * \return reordered args
   */
  FunctionModel::list_of_arguments_t reorderArgumentsByParticleTypes(
      const FunctionModel::list_of_arguments_t &eachargs,
      const ParticleTypes_t &_types
      );

  /** Filter arguments according to types, allowing multiples.
   *
   * If particle types is (0,1,2) and three arguments, each with a pair of types,
   * are given, then the alignment will be: (0,1), (0,2), and (1,2).
   *
   * \param args arguments to reorder
   * \param _types particle type vector
   * \return filtered list of args
   */
  FunctionModel::list_of_arguments_t filterArgumentsByParticleTypes(
      const FunctionModel::arguments_t &args,
      const ParticleTypes_t &_types
      );

  /** Combines two argument lists by sorting and making unique.
   *
   * @param firstargs first list of arguments
   * @param secondargs second list of arguments
   * @return concatenated lists
   */
  FunctionModel::arguments_t combineArguments(
      const FunctionModel::arguments_t &firstargs,
      const FunctionModel::arguments_t &secondargs);

  /** Combines two argument lists by concatenation.
   *
   * @param firstargs first list of arguments
   * @param secondargs second list of arguments
   * @return concatenated lists
   */
  FunctionModel::arguments_t concatenateArguments(
      const FunctionModel::arguments_t &firstargs,
      const FunctionModel::arguments_t &secondargs);

  /** Combines two argument lists by concatenation.
   *
   * @param firstlistargs first list of argument tuples
   * @param secondlistargs second list of argument tuples
   * @return concatenated lists
   */
  FunctionModel::list_of_arguments_t concatenateListOfArguments(
      const FunctionModel::list_of_arguments_t &firstlistargs,
      const FunctionModel::list_of_arguments_t &secondlistargs);

}; /* namespace Extractors */


#endif /* TRAININGDATA_EXTRACTORS_HPP_ */