Scippy

SCIP

Solving Constraint Integer Programs

heur_fuzzyround.c
Go to the documentation of this file.
1/* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */
2/* */
3/* This file is part of the program and library */
4/* SCIP --- Solving Constraint Integer Programs */
5/* */
6/* Copyright (c) 2002-2024 Zuse Institute Berlin (ZIB) */
7/* */
8/* Licensed under the Apache License, Version 2.0 (the "License"); */
9/* you may not use this file except in compliance with the License. */
10/* You may obtain a copy of the License at */
11/* */
12/* http://www.apache.org/licenses/LICENSE-2.0 */
13/* */
14/* Unless required by applicable law or agreed to in writing, software */
15/* distributed under the License is distributed on an "AS IS" BASIS, */
16/* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. */
17/* See the License for the specific language governing permissions and */
18/* limitations under the License. */
19/* */
20/* You should have received a copy of the Apache-2.0 license */
21/* along with SCIP; see the file LICENSE. If not visit scipopt.org. */
22/* */
23/* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */
24
25/**@file heur_fuzzyround.c
26 * @brief primal heuristic that constructs a feasible solution from the lp-relaxation. Round only on the state-variables (binvars)
27 * and then reconstruct the rest of the variables accordingly.
28 * @author Leon Eifler
29 */
30/*---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8----+----9----+----0----+----1----+----2*/
31
32#include <assert.h>
33#include <string.h>
34
35#include "heur_fuzzyround.h"
36
37#include "probdata_cyc.h"
38#include "scip/cons_and.h"
39
40#define HEUR_NAME "fuzzyround"
41#define HEUR_DESC "primal heuristic that constructs a feasible solution from the lp-relaxation"
42#define HEUR_DISPCHAR '&'
43#define HEUR_PRIORITY 1000
44#define HEUR_FREQ 1
45#define HEUR_FREQOFS 0
46#define HEUR_MAXDEPTH -1
47#define HEUR_TIMING SCIP_HEURTIMING_AFTERNODE
48#define HEUR_USESSUBSCIP FALSE /**< does the heuristic use a secondary SCIP instance? */
49
50/*
51 * Local methods
52 */
53
54/** execution method of primal heuristic */
55static
56SCIP_DECL_HEUREXEC(heurExecFuzzyround)
57{ /*lint --e{715}*/
58 SCIP_VAR*** binvars;
59 SCIP_SOL* sol;
60 SCIP_Real** clustering;
61 SCIP_Real maxlpval;
62 SCIP_Bool feasible = FALSE;
63 int* binsincluster;
64 int nbins;
65 int ncluster;
66 int i;
67 int k;
68 int maxcluster;
69
70 assert(heur != NULL);
71 assert(strcmp(SCIPheurGetName(heur), HEUR_NAME) == 0);
72 assert(scip != NULL);
73 assert(result != NULL);
74
75 *result = SCIP_DIDNOTRUN;
76
77 /* only call heuristic, if an optimal LP solution is at hand */
79 return SCIP_OKAY;
80
81 /* only call separator, if there are fractional variables */
82 if( SCIPgetNLPBranchCands(scip) == 0 )
83 return SCIP_OKAY;
84
85 nbins = SCIPcycGetNBins(scip);
86 ncluster = SCIPcycGetNCluster(scip);
87 assert(nbins > 0);
88 assert(ncluster > 0 && ncluster <= nbins);
89
90 binvars = SCIPcycGetBinvars(scip);
91 assert(binvars != NULL);
92
93 /* allocate memory */
94 SCIP_CALL( SCIPallocClearBufferArray(scip, &clustering , nbins) );
95 SCIP_CALL( SCIPallocClearBufferArray(scip, &binsincluster, ncluster) );
96
97 for( i = 0; i < nbins; ++i )
98 {
99 SCIP_CALL( SCIPallocClearBufferArray(scip, &clustering[i], ncluster) ); /*lint !e866*/
100 }
101
102 /* for each bin, set the assignment with the highest lp-value to 1, the rest to 0 */
103 for( i = 0; i < nbins; ++i )
104 {
105 assert(NULL != binvars[i]);
106
107 maxlpval = 0;
108 maxcluster = -1;
109
110 for (k = 0; k < ncluster; ++k)
111 {
112 assert(NULL != binvars[i][k]);
113 if( SCIPisGT(scip, SCIPvarGetLPSol(binvars[i][k]), maxlpval) )
114 {
115 maxlpval = SCIPvarGetLPSol(binvars[i][k]);
116 maxcluster = k;
117 binsincluster[k]++;
118 }
119 else if( SCIPisEQ(scip, SCIPvarGetLPSol(binvars[i][k]), maxlpval) && maxcluster != -1
120 && binsincluster[maxcluster] > binsincluster[k] )
121 {
122 binsincluster[maxcluster]--;
123 binsincluster[k]++;
124 maxcluster = k;
125 }
126 }
127
128 assert(maxcluster >= 0);
129
130 clustering[i][maxcluster] = 1.0;
131 }
132
133 assert(isPartition(scip, clustering, nbins, ncluster));
134
135 SCIP_CALL( SCIPcreateSol(scip, &sol, heur) );
136 SCIP_CALL( assignVars(scip, sol, clustering, nbins, ncluster) );
137 SCIP_CALL( SCIPtrySolFree(scip, &sol, FALSE, TRUE, TRUE, TRUE, TRUE, &feasible) );
138
139 if( feasible )
140 *result = SCIP_FOUNDSOL;
141 else
142 *result = SCIP_DIDNOTFIND;
143
144 /* free allocated memory */
145 for( i = 0; i < nbins; ++i )
146 {
147 SCIPfreeBufferArray(scip, &clustering[i]);
148 }
149 SCIPfreeBufferArray(scip, &clustering);
150 SCIPfreeBufferArray(scip, &binsincluster);
151
152 return SCIP_OKAY;
153}
154
155/*
156 * primal heuristic specific interface methods
157 */
158
159/** creates the oneopt primal heuristic and includes it in SCIP */
161 SCIP* scip /**< SCIP data structure */
162 )
163{
164 SCIP_HEUR* heur;
165
166 /* include primal heuristic */
169 HEUR_MAXDEPTH, HEUR_TIMING, HEUR_USESSUBSCIP, heurExecFuzzyround, NULL) );
170
171 assert(heur != NULL);
172
173 return SCIP_OKAY;
174}
Constraint handler for AND constraints, .
#define NULL
Definition: def.h:267
#define SCIP_Bool
Definition: def.h:91
#define SCIP_Real
Definition: def.h:173
#define TRUE
Definition: def.h:93
#define FALSE
Definition: def.h:94
#define SCIP_CALL(x)
Definition: def.h:374
int SCIPgetNLPBranchCands(SCIP *scip)
Definition: scip_branch.c:428
SCIP_RETCODE SCIPincludeHeurBasic(SCIP *scip, SCIP_HEUR **heur, const char *name, const char *desc, char dispchar, int priority, int freq, int freqofs, int maxdepth, SCIP_HEURTIMING timingmask, SCIP_Bool usessubscip, SCIP_DECL_HEUREXEC((*heurexec)), SCIP_HEURDATA *heurdata)
Definition: scip_heur.c:117
const char * SCIPheurGetName(SCIP_HEUR *heur)
Definition: heur.c:1453
SCIP_LPSOLSTAT SCIPgetLPSolstat(SCIP *scip)
Definition: scip_lp.c:168
#define SCIPallocClearBufferArray(scip, ptr, num)
Definition: scip_mem.h:126
#define SCIPfreeBufferArray(scip, ptr)
Definition: scip_mem.h:136
SCIP_RETCODE SCIPcreateSol(SCIP *scip, SCIP_SOL **sol, SCIP_HEUR *heur)
Definition: scip_sol.c:184
SCIP_RETCODE SCIPtrySolFree(SCIP *scip, SCIP_SOL **sol, SCIP_Bool printreason, SCIP_Bool completely, SCIP_Bool checkbounds, SCIP_Bool checkintegrality, SCIP_Bool checklprows, SCIP_Bool *stored)
Definition: scip_sol.c:3050
SCIP_Bool SCIPisGT(SCIP *scip, SCIP_Real val1, SCIP_Real val2)
SCIP_Bool SCIPisEQ(SCIP *scip, SCIP_Real val1, SCIP_Real val2)
SCIP_Real SCIPvarGetLPSol(SCIP_VAR *var)
Definition: var.c:18452
#define HEUR_TIMING
#define HEUR_FREQOFS
#define HEUR_DESC
#define HEUR_DISPCHAR
#define HEUR_MAXDEPTH
#define HEUR_PRIORITY
#define HEUR_NAME
SCIP_RETCODE SCIPincludeHeurFuzzyround(SCIP *scip)
#define HEUR_FREQ
static SCIP_DECL_HEUREXEC(heurExecFuzzyround)
#define HEUR_USESSUBSCIP
primal heuristic that constructs a feasible solution from the lp-relaxation. Round only on the state-...
SCIP_RETCODE assignVars(SCIP *scip, SCIP_SOL *sol, SCIP_Real **clustering, int nbins, int ncluster)
Definition: probdata_cyc.c:88
int SCIPcycGetNBins(SCIP *scip)
int SCIPcycGetNCluster(SCIP *scip)
SCIP_VAR *** SCIPcycGetBinvars(SCIP *scip)
SCIP_Bool isPartition(SCIP *scip, SCIP_Real **solclustering, int nbins, int ncluster)
Definition: probdata_cyc.c:57
problem data for cycle clustering problem
@ SCIP_LPSOLSTAT_OPTIMAL
Definition: type_lp.h:43
@ SCIP_DIDNOTRUN
Definition: type_result.h:42
@ SCIP_DIDNOTFIND
Definition: type_result.h:44
@ SCIP_FOUNDSOL
Definition: type_result.h:56
@ SCIP_OKAY
Definition: type_retcode.h:42
enum SCIP_Retcode SCIP_RETCODE
Definition: type_retcode.h:63