Match

Defines a match object used to build and refine matches between Molecule type objects. Matches are build by matching pairs of Particles from each Molecule.

[ GAMB | Source | Keywords | Summary | Ancestors | All Members | Descendants ]

Quick Index

AUTHORS

CHANGES LOG

GOALS

USAGE

Class Summary

Back to the top of Match

AUTHORS

Copyright: SAMBA group, Tel-Aviv Univ. Israel, 1997.

Back to the top of Match

CHANGES LOG

Back to the top of Match

GOALS

Special attention should be given to the add method. This method lets the user add a pair to the match. Besides specifying the indices of the Particles within the Molcules the add method recieves a score parameter and a priority parameter. The score parameter specifies how much the Particle pair contributes to the score. The priority tells the add method how "important" the pair is. If a different pair is added, one that clashes with the first pair (i.e. the particles overlap) add has to decide which pair to bump off). The descision is made using the highest priority. An example of a priority function would the inverse distance. In this case add will prefer the closer pair even if the pair with the longer distance scores better.

Becasue of the difficulty in determinig how a match should be constucted and built the non-public members of this class were left protected. This allows a user to write Match descendants that best fit his needs.

Back to the top of Match

USAGE

    Match M;
    for (int i=0; i<1000; i++) {
      for (int j=0; j<200; j++) {
        // randomly pick pair
        unsigned int modelParticle = model.size()*drand48();
	unsigned int sceneParticle = scene.size()*drand48();
	// score is pair distance inverse. 
	float score = (1 / (1+(model[modelParticle] | scene[sceneParticle])));
	// add pair according to score
	M.add(modelParticle, sceneParticle, score ,score);
      } // j
      M.calculateBestFit(model, scene);  // every 200 random pairs recalc
      cout << M.rigidTrans() << '\n';    // rig. trans. and print it.
    } // i

    int matchingParticle;
    for(int i=0; i<model.size(); i++)
      if ((matchingParticle = M.sceneParticle(i)) >= 0)
        cout << i << '\t' << matchingParticle << '\n';

Comment: Generally, it is expected (By me at least) that if a very large linear match infact exists between model and scene (e.g. try model=scene) the calculated rigid transformation will probablistically tend to such a match.

Back to the top of Match

class ParticleMatch ;

  class ParticleMatch
  {
  public:
    ParticleMatch();
    ParticleMatch(unsigned int model, unsigned int scene, 
		  float sc, float pri);

Back to the top of Match

Match();

  Match();

Back to the top of Match

Match(const RigidTrans3& rt);

  Match(const RigidTrans3& rt);

Back to the top of Match

Match(const Match& m);

  Match(const Match& m);

Back to the top of Match

bool add(const unsigned int modelParticle, const unsigned int sceneParticle, const float score, const float priority);

  bool add(const unsigned int modelParticle, const unsigned int sceneParticle, 
           const float score, 
           const float priority);

Back to the top of Match

void add(const unsigned int modelParticle, const unsigned int sceneParticle, const float score );

  void add(const unsigned int modelParticle, const unsigned int sceneParticle, 
           const float score = 1.0);

Back to the top of Match

void add(const ParticleMatch& pm);

  void add(const ParticleMatch& pm);

Back to the top of Match

float calculateTotalScore();

  float calculateTotalScore();

Back to the top of Match

short size() const;

  short size() const;

Back to the top of Match

float rmsd() const;

  float rmsd() const;

Back to the top of Match

float totalScore() const;

  float totalScore() const;

Back to the top of Match

const RigidTrans3& rigidTrans() const;

  const RigidTrans3& rigidTrans() const;

Back to the top of Match

int sceneParticle(const unsigned int model) const;

  int sceneParticle(const unsigned int model) const;

Back to the top of Match

int modelParticle(const unsigned int scene) const;

  int modelParticle(const unsigned int scene) const;

Back to the top of Match

const ParticleMatch& operator[](const unsigned int index) const ;

  const ParticleMatch& operator[](const unsigned int index) const                      
;

Function is currently defined inline.

Back to the top of Match

ParticleMatch& operator[](const unsigned int index) ;

  ParticleMatch& operator[](const unsigned int index)                      
;

Function is currently defined inline.

Back to the top of Match

template< class TMolecule> void calculateBestFit(const TMolecule& model, const TMolecule& scene);

  template< class TMolecule>
  void calculateBestFit(const TMolecule& model, const TMolecule& scene);  

Back to the top of Match

template< class TMolecule> float calculateLeastSquaresError(const TMolecule& model, const TMolecule& scene);

Calculates the least-squares error between the two given TMolecules. A negative value is returned in case of an error.
TMolecule is assumed to be a collection of elements with 3D coordinates. Thus, TMolecule should support operator[] that returns an element with a cast Vector3() opertator. Examples for TMolecule are: vector<Vector3> and Molecule<Atom>.
Author: Oranit Dror (oranit@tau.ac.il)

  template< class TMolecule>
  static float calculateLeastSquaresError(const TMolecule& model, const TMolecule& scene);

Back to the top of Match

template< class TMolecule> float calculateRMSD(const TMolecule& model, const TMolecule& scene);

Calculates the RMSD between the two given TMolecules. A negative is be returned in case of an error.
TMolecule is assumed to be a collection of elements with 3D coordinates. Thus, TMolecule should support operator[] that returns an element with a cast Vector3() opertator. Examples for TMolecule are: vector<Vector3> and Molecule<Atom>.
Author: Oranit Dror (oranit@tau.ac.il)

  template< class TMolecule>
  static float calculateRMSD(const TMolecule& model, const TMolecule& scene);

Back to the top of Match

template< class TComperator> void sortMatchList(const TComperator& comperator);

Sort the match list according to the given Comperator object. The Comperator object is a Binary Predicate that compares two objects, returning true if the first precedes the second. This predicate must satisfy the standard mathematical definition of a strict weak ordering. The precise requirements are stated in the STL StrictWeakOrdering documentation, but what they roughly mean is that the predicate has to behave the way that "less than" behaves: if a is less than b then b is not less than a, if a is less than b and b is less than c then a is less than c, and so on. Author: Oranit Dror (oranit@tau.ac.il)

  template< class TComperator>
  void sortMatchList(const TComperator& comperator);

Back to the top of Match

void clear() ;

Erase the particle pairs of the match list

  void clear()                         
;

Function is currently defined inline.

Back to the top of Match

All Members

Back to the top of Match

Ancestors

Back to the top of Match

Descendants

Back to the top of Match