Detailed Description

problem data for cycle clustering problem

Author: Leon Eifler

This file implements the problem data for the cycle clustering problem.

The problem data contains original transition matrix, the scaling parameter that appears in the objective function, and all variables that appear in the problem.

Definition in file probdata_cyc.c.

#include "probdata_cyc.h"
#include "scip/cons_nonlinear.h"
#include "scip/cons_linear.h"
#include "scip/cons_logicor.h"
#include "scip/var.h"
#include <assert.h>

Go to the source code of this file.

Functions
SCIP_Bool	isPartition (SCIP scip, SCIP_Real *solclustering, int nbins, int ncluster)

SCIP_RETCODE	assignVars (SCIP scip, SCIP_SOL sol, SCIP_Real **clustering, int nbins, int ncluster)

int	phi (int k, int ncluster)

int	phiinv (int k, int ncluster)

static SCIP_RETCODE	createVariables (SCIP scip, SCIP_PROBDATA probdata)

static SCIP_RETCODE	createProbSimplifiedTest (SCIP scip, SCIP_PROBDATA probdata)

static SCIP_RETCODE	createProbSimplified (SCIP scip, SCIP_PROBDATA probdata)

static SCIP_RETCODE	createProbQP (SCIP scip, SCIP_PROBDATA probdata)

static SCIP_RETCODE	createProbOnlyEdge (SCIP scip, SCIP_PROBDATA probdata)

static	SCIP_DECL_PROBTRANS (probtransCyc)

static	SCIP_DECL_PROBDELORIG (probdelorigCyc)

static	SCIP_DECL_PROBDELTRANS (probdeltransCyc)

static	SCIP_DECL_PROBCOPY (probcopyCyc)

SCIP_RETCODE	SCIPcreateProbCyc (SCIP scip, const char name, int nbins, int ncluster, SCIP_Real **cmatrix)

SCIP_Real **	SCIPcycGetCmatrix (SCIP *scip)

int	SCIPcycGetNBins (SCIP *scip)

int	SCIPcycGetNCluster (SCIP *scip)

SCIP_VAR ***	SCIPcycGetBinvars (SCIP *scip)

SCIP_Real	SCIPcycGetScale (SCIP *scip)

SCIP_VAR ****	SCIPcycGetEdgevars (SCIP *scip)

SCIP_VAR *	getEdgevar (SCIP_VAR ****edgevars, int state1, int state2, int direction)

SCIP_Bool	edgesExist (SCIP_VAR ***edgevars, int states, int nstates)

SCIP_DIGRAPH *	SCIPcycGetEdgeGraph (SCIP *scip)

SCIP_RETCODE	SCIPcycPrintSolutionValues (SCIP scip, SCIP_SOL sol)

Function Documentation

◆ isPartition()

SCIP_Bool isPartition	(	SCIP *	scip,
		SCIP_Real **	solclustering,
		int	nbins,
		int	ncluster
	)

Check if the clustering has exactly one state in every cluster.

Parameters

scip	SCIP data structure
solclustering	matrix with the clustering
nbins	the number of bins
ncluster	the number of clusters

Definition at line 48 of file probdata_cyc.c.

References FALSE, SCIP_Real, SCIPisEQ(), SCIPisIntegral(), SCIPisZero(), and TRUE.

Referenced by createSwitchSolution(), runCyckerlin(), SCIP_DECL_HEUREXEC(), and switchNext().

◆ assignVars()

SCIP_RETCODE assignVars	(	SCIP *	scip,
		SCIP_SOL *	sol,
		SCIP_Real **	clustering,
		int	nbins,
		int	ncluster
	)

Assign the variables in scip according to the found clustering.

Parameters

scip	SCIP data structure
sol	the SCIP solution
clustering	the matrix with the clusterassignment
nbins	the number of bins
ncluster	the number of cluster

Definition at line 79 of file probdata_cyc.c.

References NULL, phi(), SCIP_CALL, SCIP_OKAY, SCIP_VARSTATUS_MULTAGGR, SCIPcycGetBinvars(), SCIPcycGetEdgevars(), SCIPisEQ(), SCIPisGE(), SCIPisIntegral(), SCIPisLE(), SCIPsetSolVal(), SCIPvarGetLbGlobal(), SCIPvarGetStatus(), SCIPvarGetTransVar(), SCIPvarGetUbGlobal(), and SCIPvarIsTransformed().

Referenced by createSwitchSolution(), SCIP_DECL_HEUREXEC(), and SCIPcycAddIncompleteSol().

◆ phi()

int phi	(	int	k,
		int	ncluster
	)

function that returns the successive cluster along the cycle

Parameters

k	the cluster
ncluster	the number of clusters

Definition at line 172 of file probdata_cyc.c.

Referenced by assignVars(), createProbQP(), createProbSimplified(), and SCIPcycPrintSolutionValues().

◆ phiinv()

int phiinv	(	int	k,
		int	ncluster
	)

function that returns the predecessor-cluster along the cycle

Parameters

k	the cluster
ncluster	the number of clusters

Definition at line 184 of file probdata_cyc.c.

Referenced by createProbSimplified(), and getTempObj().

◆ createVariables()

static SCIP_RETCODE createVariables	(	SCIP *	scip,
		SCIP_PROBDATA *	probdata
	)

static

creates all the variables for the problem. The constraints are added later, depending on the model that is used

Parameters

scip	SCIP data Structure
probdata	the problem data

Definition at line 200 of file probdata_cyc.c.

References NULL, SCIP_CALL, SCIP_MAXSTRLEN, SCIP_OBJSENSE_MAXIMIZE, SCIP_OKAY, SCIP_VARTYPE_BINARY, SCIPaddVar(), SCIPallocBlockMemoryArray, SCIPallocClearBlockMemoryArray, SCIPcreateVarBasic(), SCIPdigraphAddArc(), SCIPisZero(), SCIPsetObjsense(), and SCIPsnprintf().

Referenced by SCIPcreateProbCyc().

◆ createProbSimplifiedTest()

static SCIP_RETCODE createProbSimplifiedTest	(	SCIP *	scip,
		SCIP_PROBDATA *	probdata
	)

static

create the problem without variable amount of clusters, use simpler non-facet-defining inequalities

Parameters

scip	SCIP Data Structure
probdata	the problem data

Definition at line 269 of file probdata_cyc.c.

References FALSE, NULL, SCIP_CALL, SCIP_MAXSTRLEN, SCIP_OKAY, SCIP_Real, SCIPaddCoefLinear(), SCIPaddCoefLogicor(), SCIPaddCoefSetppc(), SCIPaddCons(), SCIPchgVarObj(), SCIPcreateConsBasicLinear(), SCIPcreateConsBasicLogicor(), SCIPcreateConsLinear(), SCIPcreateConsSetpart(), SCIPgetRealParam(), SCIPinfinity(), SCIPinfoMessage(), SCIPreleaseCons(), SCIPsnprintf(), and TRUE.

Referenced by SCIPcreateProbCyc().

◆ createProbSimplified()

static SCIP_RETCODE createProbSimplified	(	SCIP *	scip,
		SCIP_PROBDATA *	probdata
	)

static

create the problem without variable amount of clusters, using three edge-variables for each pair of states. This is the tested default version.

Parameters

scip	SCIP Data Structure
probdata	The problem data

Definition at line 411 of file probdata_cyc.c.

References FALSE, NULL, phi(), phiinv(), SCIP_CALL, SCIP_MAXSTRLEN, SCIP_OKAY, SCIP_Real, SCIP_VERBLEVEL_NORMAL, SCIPaddCoefLinear(), SCIPaddCoefLogicor(), SCIPaddCoefSetppc(), SCIPaddCons(), SCIPaddOrigObjoffset(), SCIPchgVarObj(), SCIPcreateConsBasicLinear(), SCIPcreateConsBasicLogicor(), SCIPcreateConsLinear(), SCIPcreateConsSetpart(), SCIPgetRealParam(), SCIPinfinity(), SCIPisZero(), SCIPreleaseCons(), SCIPsnprintf(), SCIPverbMessage(), and TRUE.

Referenced by SCIPcreateProbCyc().

◆ createProbQP()

static SCIP_RETCODE createProbQP	(	SCIP *	scip,
		SCIP_PROBDATA *	probdata
	)

static

create the problem without variable amount of clusters, using quadratic formulations. This is inferior to the simplified variant. Only useful for comparing relaxations.

Parameters