TSPCostC -

Developer Documentation

Centre for Vision, Speech & Signal Processing

Comments:
------------------------------------------------------------------------- ************************************************************************* ------------------------------------------------------------------------- ------------------------------------------------------------------------- ------------------------------------------------------------------------- A simple cost to test various optimizers. The cost is the travelling salesman problem. The cities are randomly distributed within the unit cube. The problem is completely determined by the coefficients E(i,j,a,b) And we seek a solution with the i-th visit being to city a-th. The trip must start from (0,0,0) and finish in (1,1,1) after visiting all cities. For (i != j=/-1 ) the only cost is a penalty to avoid double visits This will not affect an assignment algorithm. nlabs must be equal nobjs It appears that the TSP problem with Grad Ass can flip flop since the global best path between 2 points may not be the local best path, surprisingly SA does OK on this. -------------------------------------------------------------------------

Parent Classes:

SimplePairwiseCostC

Variables:

SArray1dC towns;

Methods:

TSPCostC(TraceC t = TraceC(),int _nobjs = 10,int idum = -1)
~TSPCostC(): Constructor
double Energy(int i,int j,int a,int b): double Energy(const HardAssignC & r) { return SimplePairwiseCostC::Energy(r); }

#include "amma/BaseCost.hh"

double Energy(int i,int j,int a,int b): return the pairwise energy Thes functions are taken care of now for you --------------------------------------------
double Energy(const HardAssignC & r)
double Energy(const SoftAssignC & r): return the energy of the overall labelling It is assumed that this is slower and to be used sparingly MFTCostC stuff --------------
void GetMFTSupport(SoftAssignC & q,int i,const SoftAssignC & work): returns the support for one object
SoftAssignC GetMFTSupport(const SoftAssignC & work): returns the support GradAssCostC stuff ------------------
void GetGradAssSupport(SoftAssignC & q,int i,const SoftAssignC & work): returns the support for one object
SoftAssignC GetGradAssSupport(const SoftAssignC & work): returns the support SACostC stuff -------------
double GetEnergy(int i,int a,const HardAssignC & r): get the energy for object i with label a, or at least the part of the energy affected by changing labels on object i only
RealSArray1dC LoadEnergy(int i,const HardAssignC & r): Load up the array work with the energies for each label of object i given global assignment r
void TrapUndefinedVirtual(const char * fname): the crude but efective way to catch virtual functions that have not been overridden
double Energy(int i,int j,int a,int b): return the pairwise energy
double Energy(const HardAssignC & r): return the energy of the overall labelling It is assumed that this is slower and to be used sparingly
double Energy(const SoftAssignC & r): return the energy of the overall labelling It is assumed that this is slower and to be used sparingly Mean Field Theory -----------------
void GetMFTSupport(SoftAssignC & q,int i,const SoftAssignC & work): returns the support for one object
SoftAssignC GetMFTSupport(const SoftAssignC & work): returns the support Graduated assignment --------------------
void GetGradAssSupport(SoftAssignC & q,int i,const SoftAssignC & work): returns the support for one object
SoftAssignC GetGradAssSupport(const SoftAssignC & work): returns the support Simulated Annealing -------------------
double GetEnergy(int i,int a,const HardAssignC & r): get the energy for object i with label a, or at least the part of the energy affected by changing labels on object i only
RealSArray1dC LoadEnergy(int i,const HardAssignC & r): Load up the array work with the energies for each label of object i given global assignment r

#include "amma/Trace.hh"

Inline TraceC(int tl = NoTrace,ostream & os = cout): Constructor
Inline ~TraceC(): Destructor
Inline TraceC(const TraceC & t): Copy Constructor

Programmer:Andrew Stoddart, Documentation by CxxDoc: Tue Mar 20 10:49:27 2001

TSPCostC

include	"amma/TSPCost.hh"
User Level:	Default
Library:	CombTest
Example:	doNumClassComb.cc
Section:	default.Andrew Stoddart
In Scope:	std