Scippy

SCIP

Solving Constraint Integer Programs

pricer_vrp.cpp
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 pricer_vrp.cpp
26 * @brief VRP pricer plugin
27 * @author Andreas Bley
28 * @author Marc Pfetsch
29 */
30
31#include "pricer_vrp.h"
32#include "pqueue.h"
33
34#include <iostream>
35#include <map>
36#include <vector>
37
38#include "scip/cons_linear.h"
39
40using namespace std;
41using namespace scip;
42
43
44
45
46/** Constructs the pricer object with the data needed
47 *
48 * An alternative is to have a problem data class which allows to access the data.
49 */
51 SCIP* scip, /**< SCIP pointer */
52 const char* p_name, /**< name of pricer */
53 const int p_num_nodes, /**< number of nodes */
54 const int p_capacity, /**< vehicle capacity */
55 const vector< int >& p_demand, /**< demand array */
56 const vector< vector<int> >& p_distance, /**< matrix of distances */
57 const vector< vector<SCIP_VAR*> >& p_arc_var, /**< matrix of arc variables */
58 const vector< vector<SCIP_CONS*> >& p_arc_con, /**< matrix of arc constraints */
59 const vector<SCIP_CONS* >& p_part_con /**< array of partitioning constraints */
60 ):
61 ObjPricer(scip, p_name, "Finds tour with negative reduced cost.", 0, TRUE),
62 _num_nodes(p_num_nodes),
63 _capacity(p_capacity),
64 _demand(p_demand),
65 _distance(p_distance),
66 _arc_var(p_arc_var),
67 _arc_con(p_arc_con),
68 _part_con(p_part_con)
69{}
70
71
72/** Destructs the pricer object. */
74{}
75
76
77/** initialization method of variable pricer (called after problem was transformed)
78 *
79 * Because SCIP transformes the original problem in preprocessing, we need to get the references to
80 * the variables and constraints in the transformed problem from the references in the original
81 * problem.
82 */
83SCIP_DECL_PRICERINIT(ObjPricerVRP::scip_init)
84{
85 for (int i = 0; i < num_nodes(); ++i)
86 {
87 for (int j = 0; j < i; ++j)
88 {
89 SCIP_CALL( SCIPgetTransformedVar(scip, _arc_var[i][j], &_arc_var[i][j]) ); /*lint !e732 !e747*/
90 SCIP_CALL( SCIPgetTransformedCons(scip, _arc_con[i][j], &_arc_con[i][j]) ); /*lint !e732 !e747*/
91 }
92 }
93 for (int i = 1; i < num_nodes(); ++i)
94 {
95 SCIP_CALL( SCIPgetTransformedCons(scip, _part_con[i], &_part_con[i]) ); /*lint !e732 !e747*/
96 }
97
98 return SCIP_OKAY;
99} /*lint !e715*/
100
101
102/** perform pricing
103 *
104 * @todo compute shortest length restricted tour w.r.t. duals
105 */
107 SCIP* scip, /**< SCIP data structure */
108 bool isfarkas /**< whether we perform Farkas pricing */
109 ) const
110{
111 /* allocate array for reduced costs */
112 vector< vector<SCIP_Real> > red_length(num_nodes()); /*lint !e732 !e747*/
113 for (int i = 0; i < num_nodes(); ++i)
114 red_length[i].resize(i, 0.0); /*lint !e732 !e747*/
115
116 /* compute reduced-cost arc lengths store only lower triangualar matrix, i.e., red_length[i][j] only for i > j */
117 if ( isfarkas )
118 {
119 for (int i = 0; i < num_nodes(); ++i)
120 {
121 assert( i == 0 || part_con(i) != 0 );
122 for (int j = 0; j < i; ++j)
123 {
124 SCIP_Real r = 0.0;
125 assert( arc_con(i,j) != 0 );
126
128 if ( j != 0 )
130 if ( i != 0 )
132 red_length[i][j] = r; /*lint !e732 !e747*/
133 }
134 }
135 }
136 else
137 {
138 for (int i = 0; i < num_nodes(); ++i)
139 {
140 assert( i == 0 || part_con(i) != 0 );
141 for (int j = 0; j < i; ++j)
142 {
143 SCIP_Real r = 0.0;
144 assert( arc_con(i,j) != 0 );
145
147 if ( j != 0 )
148 r -= 0.5 * SCIPgetDualsolLinear(scip, part_con(j));
149 if ( i != 0 )
150 r -= 0.5 * SCIPgetDualsolLinear(scip, part_con(i));
151 red_length[i][j] = r; /*lint !e732 !e747*/
152 }
153 }
154 }
155
156#ifdef SCIP_OUTPUT
157 if ( isfarkas )
158 {
159 SCIPinfoMessage(scip, NULL, "dual ray solution:\n");
160 for (int i = 0; i < num_nodes(); ++i)
161 {
162 for (int j = 0; j < i; ++j)
163 SCIPinfoMessage(scip, NULL, "arc_%d_%d: %g\n", i, j, SCIPgetDualfarkasLinear(scip, arc_con(i,j)));
164 }
165
166 for (int i = 1; i < num_nodes(); ++i)
167 SCIPinfoMessage(scip, NULL, "part_%d: %g\n", i, SCIPgetDualfarkasLinear(scip, part_con(i)));
168
169 for (int i = 0; i < num_nodes(); ++i)
170 {
171 for (int j = 0; j < i; ++j)
172 SCIPinfoMessage(scip, NULL, "length_%d_%d: %g\n", i, j, red_length[i][j]);
173 }
174 }
175 else
176 {
177 SCIPinfoMessage(scip, NULL, "dual solution:\n");
178 for (int i = 0; i < num_nodes(); ++i)
179 {
180 for (int j = 0; j < i; ++j)
181 SCIPinfoMessage(scip, NULL, "arc_%d_%d: %g\n", i, j, SCIPgetDualsolLinear(scip, arc_con(i,j)));
182 }
183
184 for (int i = 1; i < num_nodes(); ++i)
185 SCIPinfoMessage(scip, NULL, "part_%d: %g\n", i, SCIPgetDualsolLinear(scip, part_con(i)));
186
187 for (int i = 0; i < num_nodes(); ++i)
188 {
189 for (int j = 0; j < i; ++j)
190 SCIPinfoMessage(scip, NULL, "length_%d_%d: %g\n", i, j, red_length[i][j]);
191 }
192 }
193#endif
194
195 /* compute shortest length restricted tour w.r.t. reduced-cost arc length */
196 list<int> tour;
197 SCIP_Real reduced_cost = find_shortest_tour(red_length, tour);
198
199 /* add tour variable */
200 if ( SCIPisNegative(scip, reduced_cost) )
201 {
202 return add_tour_variable(scip, tour);
203 }
204
205#ifdef SCIP_OUTPUT
206 SCIP_CALL( SCIPwriteTransProblem(scip, "vrp.lp", "lp", FALSE) );
207#endif
208
209 return SCIP_OKAY;
210}
211
212
213
214/** Pricing of additional variables if LP is feasible.
215 *
216 * - get the values of the dual variables you need
217 * - construct the reduced-cost arc lengths from these values
218 * - find the shortest admissible tour with respect to these lengths
219 * - if this tour has negative reduced cost, add it to the LP
220 *
221 * possible return values for *result:
222 * - SCIP_SUCCESS : at least one improving variable was found, or it is ensured that no such variable exists
223 * - SCIP_DIDNOTRUN : the pricing process was aborted by the pricer, there is no guarantee that the current LP solution is optimal
224 */
225SCIP_DECL_PRICERREDCOST(ObjPricerVRP::scip_redcost)
226{
227 SCIPdebugMsg(scip, "call scip_redcost ...\n");
228
229 /* set result pointer, see above */
230 *result = SCIP_SUCCESS;
231
232 /* call pricing routine */
233 SCIP_CALL( pricing(scip, false) );
234
235 return SCIP_OKAY;
236} /*lint !e715*/
237
238
239/** Pricing of additional variables if LP is infeasible.
240 *
241 * - get the values of the dual Farks multipliers you need
242 * - construct the reduced-cost arc lengths from these values
243 * - find the shortest admissible tour with respect to these lengths
244 * - if this tour has negative reduced cost, add it to the LP
245 */
246SCIP_DECL_PRICERFARKAS(ObjPricerVRP::scip_farkas)
247{
248 SCIPdebugMsg(scip, "call scip_farkas ...\n");
249
250 /* call pricing routine */
251 SCIP_CALL( pricing(scip, true) );
252
253 return SCIP_OKAY;
254} /*lint !e715*/
255
256
257/** add tour variable to problem */
259 SCIP* scip, /**< SCIP data structure */
260 const list<int>& tour /**< list of nodes in tour */
261 ) const
262{
263 /* create meaningful variable name */
264 char tmp_name[255];
265 char var_name[255];
266 (void) SCIPsnprintf(var_name, 255, "T");
267 for (list<int>::const_iterator it = tour.begin(); it != tour.end(); ++it) /*lint !e1702*/
268 {
269 strncpy(tmp_name, var_name, 255);
270 (void) SCIPsnprintf(var_name, 255, "%s_%d", tmp_name, *it);
271 }
272 SCIPdebugMsg(scip, "new variable <%s>\n", var_name);
273
274 /* create the new variable: Use upper bound of infinity such that we do not have to care about
275 * the reduced costs of the variable in the pricing. The upper bound of 1 is implicitly satisfied
276 * due to the set partitioning constraints.
277 */
278 SCIP_VAR* var;
279 SCIP_CALL( SCIPcreateVar(scip, &var, var_name,
280 0.0, // lower bound
281 SCIPinfinity(scip), // upper bound
282 0.0, // objective
283 SCIP_VARTYPE_CONTINUOUS, // variable type
284 false, false, NULL, NULL, NULL, NULL, NULL) );
285
286 /* add new variable to the list of variables to price into LP (score: leave 1 here) */
287 SCIP_CALL( SCIPaddPricedVar(scip, var, 1.0) );
288
289 /* add coefficient into the set partition constraints */
290 for (list<int>::const_iterator it = tour.begin(); it != tour.end(); ++it) /*lint !e1702*/
291 {
292 assert( 0 <= *it && *it < num_nodes() );
293 SCIP_CALL( SCIPaddCoefLinear(scip, part_con(*it), var, 1.0) );
294 }
295
296 /* add coefficient into arc routing constraints */
297 int last = 0;
298 for (list<int>::const_iterator it = tour.begin(); it != tour.end(); ++it) /*lint !e1702*/
299 {
300 assert( 0 <= *it && *it < num_nodes() );
301 SCIP_CALL( SCIPaddCoefLinear(scip, arc_con(last, *it), var, 1.0) );
302 last = *it;
303 }
304 SCIP_CALL( SCIPaddCoefLinear(scip, arc_con(last, 0), var, 1.0 ) );
305
306 /* cleanup */
307 SCIP_CALL( SCIPreleaseVar(scip, &var) );
308
309 return SCIP_OKAY;
310}
311
312
313/** Computes a shortest admissible tour with respect to the given lengths. The function must return
314 * the computed tour via the parameter tour and the length (w.r.t. given lengths) of this tour as
315 * return parameter. The returned tour must be the ordered list of customer nodes contained in the
316 * tour (i.e., 2-5-7 for the tour 0-2-5-7-0).
317 */
318namespace
319{
320
321/* types needed for prioity queue -------------------- */
322static const SCIP_Real eps = 1e-9;
323
324struct PQUEUE_KEY
325{
326 int demand;
327 SCIP_Real length;
328
329 PQUEUE_KEY() : demand(0), length(0.0) {}
330};
331
332bool operator< (const PQUEUE_KEY& l1, const PQUEUE_KEY& l2)
333{
334 if ( l1.demand < l2.demand )
335 return true;
336 if ( l1.demand > l2.demand )
337 return false;
338 if ( l1.length < l2.length-eps )
339 return true;
340 /* not needed, since we return false anyway:
341 if ( l1.length > l2.length+eps )
342 return false;
343 */
344 return false;
345}
346
347typedef int PQUEUE_DATA; // node
348typedef pqueue<PQUEUE_KEY,PQUEUE_DATA> PQUEUE;
349typedef PQUEUE::pqueue_item PQUEUE_ITEM;
350
351
352/* types needed for dyn. programming table */
353struct NODE_TABLE_DATA
354{
355 SCIP_Real length;
356 int predecessor;
357 PQUEUE::pqueue_item queue_item;
358
359 NODE_TABLE_DATA( ) : length(0.0), predecessor(-1), queue_item( NULL ) {}
360};
361
362typedef int NODE_TABLE_KEY; // demand
363typedef std::map< NODE_TABLE_KEY, NODE_TABLE_DATA > NODE_TABLE;
364}
365
366
367/** return negative reduced cost tour (uses restricted shortest path dynamic programming algorithm)
368 *
369 * The algorithm uses the priority queue implementation in pqueue.h. SCIP's implementation of
370 * priority queues cannot be used, since it currently does not support removal of elements that are
371 * not at the top.
372 */
374 const vector< vector<SCIP_Real> >& length, /**< matrix of lengths */
375 list<int>& tour /**< list of nodes in tour */
376 ) const
377{
378 tour.clear();
379
380 SCIPdebugMessage("Enter RSP - capacity: %d\n", capacity());
381
382 /* begin algorithm */
383 PQUEUE PQ;
384 vector< NODE_TABLE > table(num_nodes()); /*lint !e732 !e747*/
385
386 /* insert root node (start at node 0) */
387 PQUEUE_KEY queue_key;
388 PQUEUE_DATA queue_data = 0;
389 PQUEUE_ITEM queue_item = PQ.insert(queue_key, queue_data);
390
391 NODE_TABLE_KEY table_key = 0;
392 NODE_TABLE_DATA table_entry;
393
394 /* run Dijkstra-like updates */
395 while ( ! PQ.empty() )
396 {
397 /* get front queue entry */
398 queue_item = PQ.top();
399 queue_key = PQ.get_key (queue_item);
400 queue_data = PQ.get_data(queue_item);
401 PQ.pop();
402
403 /* get corresponding node and node-table key */
404 const int curr_node = queue_data;
405 const SCIP_Real curr_length = queue_key.length;
406 const int curr_demand = queue_key.demand;
407
408 /* stop as soon as some negative length tour was found */
409 if ( curr_node == 0 && curr_length < -eps )
410 break;
411
412 /* stop as soon don't create multi-tours */
413 if ( curr_node == 0 && curr_demand != 0 )
414 continue;
415
416 /* update all active neighbors */
417 for (int next_node = 0; next_node < num_nodes(); ++next_node)
418 {
419 if ( next_node == curr_node )
420 continue;
421 if ( have_edge( next_node, curr_node ) == false )
422 continue;
423
424 const int next_demand = curr_demand + demand(next_node);
425
426 if ( next_demand > capacity() )
427 continue;
428
429 const SCIP_Real next_length = curr_length + ( curr_node > next_node ? /*lint !e732 !e747*/
430 length[curr_node][next_node] : /*lint !e732 !e747*/
431 length[next_node][curr_node] ); /*lint !e732 !e747*/
432
433 NODE_TABLE& next_table = table[next_node]; /*lint !e732 !e747*/
434
435 /* check if new table entry would be dominated */
436 bool skip = false;
437 list<NODE_TABLE::iterator> dominated;
438
439 for (NODE_TABLE::iterator it = next_table.begin(); it != next_table.end() && ! skip; ++it) /*lint !e1702*/
440 {
441 if ( next_demand >= it->first && next_length >= it->second.length - eps )
442 skip = true;
443
444 if ( next_demand <= it->first && next_length <= it->second.length + eps )
445 dominated.push_front( it );
446 }
447 if ( skip )
448 continue;
449
450 /* remove dominated table and queue entries */
451 for (list<NODE_TABLE::iterator>::iterator it = dominated.begin(); it != dominated.end(); ++it) /*lint !e1702*/
452 {
453 PQ.remove( (*it)->second.queue_item );
454 next_table.erase( *it );
455 }
456
457 /* insert new table and queue entry */
458 queue_key.demand = next_demand;
459 queue_key.length = next_length;
460 queue_data = next_node;
461
462 queue_item = PQ.insert(queue_key, queue_data);
463
464 table_key = next_demand;
465 table_entry.length = next_length;
466 table_entry.predecessor = curr_node;
467 table_entry.queue_item = queue_item;
468
469 next_table[table_key] = table_entry;
470
471#ifdef SCIP_OUTPUT
472 printf("new entry node = %d demand = %d length = %g pref = %d\n", next_node, next_demand, next_length, curr_node);
473#endif
474 }
475 }
476
477 SCIPdebugMessage("Done RSP DP.\n");
478
479 table_entry.predecessor = -1;
480 table_entry.length = 0;
481 int curr_node = 0;
482
483 /* find most negative tour */
484 for (NODE_TABLE::iterator it = table[0].begin(); it != table[0].end(); ++it) /*lint !e1702 !e732 !e747*/
485 {
486 if ( it->second.length < table_entry.length )
487 {
488 table_key = it->first;
489 table_entry = it->second;
490 }
491 }
492 SCIP_Real tour_length = table_entry.length;
493
494 while ( table_entry.predecessor > 0 )
495 {
496 table_key -= demand(curr_node);
497 curr_node = table_entry.predecessor;
498 tour.push_front(curr_node);
499 table_entry = table[curr_node][table_key]; /*lint !e732 !e747*/
500 }
501
502 SCIPdebugMessage("Leave RSP tour length = %g\n", tour_length);
503
504 return tour_length;
505}
SCIP_Real * r
Definition: circlepacking.c:59
int num_nodes() const
Definition: pricer_vrp.h:96
ObjPricerVRP(SCIP *scip, const char *p_name, const int p_num_nodes, const int p_capacity, const vector< int > &p_demand, const vector< vector< int > > &p_distance, const vector< vector< SCIP_VAR * > > &p_arc_var, const vector< vector< SCIP_CONS * > > &p_arc_con, const vector< SCIP_CONS * > &p_part_con)
Definition: pricer_vrp.cpp:50
virtual ~ObjPricerVRP()
Definition: pricer_vrp.cpp:73
SCIP_RETCODE pricing(SCIP *scip, bool isfarkas) const
Definition: pricer_vrp.cpp:106
int capacity() const
Definition: pricer_vrp.h:102
bool have_edge(const int i, const int j) const
Definition: pricer_vrp.h:151
SCIP_RETCODE add_tour_variable(SCIP *scip, const list< int > &tour) const
Definition: pricer_vrp.cpp:258
SCIP_CONS * arc_con(const int i, const int j) const
Definition: pricer_vrp.h:134
int demand(const int i) const
Definition: pricer_vrp.h:108
double find_shortest_tour(const vector< vector< double > > &length, list< int > &tour) const
Definition: pricer_vrp.cpp:373
SCIP_CONS * part_con(const int i) const
Definition: pricer_vrp.h:143
C++ wrapper for variable pricer.
Definition: objpricer.h:53
Constraint handler for linear constraints in their most general form, .
#define NULL
Definition: def.h:266
#define SCIP_Real
Definition: def.h:172
#define TRUE
Definition: def.h:93
#define FALSE
Definition: def.h:94
#define SCIP_CALL(x)
Definition: def.h:373
SCIP_Real SCIPgetDualsolLinear(SCIP *scip, SCIP_CONS *cons)
SCIP_RETCODE SCIPaddCoefLinear(SCIP *scip, SCIP_CONS *cons, SCIP_VAR *var, SCIP_Real val)
SCIP_Real SCIPgetDualfarkasLinear(SCIP *scip, SCIP_CONS *cons)
SCIP_RETCODE SCIPaddPricedVar(SCIP *scip, SCIP_VAR *var, SCIP_Real score)
Definition: scip_prob.c:1733
SCIP_RETCODE SCIPwriteTransProblem(SCIP *scip, const char *filename, const char *extension, SCIP_Bool genericnames)
Definition: scip_prob.c:648
void SCIPinfoMessage(SCIP *scip, FILE *file, const char *formatstr,...)
Definition: scip_message.c:208
#define SCIPdebugMsg
Definition: scip_message.h:78
SCIP_RETCODE SCIPgetTransformedCons(SCIP *scip, SCIP_CONS *cons, SCIP_CONS **transcons)
Definition: scip_cons.c:1675
SCIP_Real SCIPinfinity(SCIP *scip)
SCIP_Bool SCIPisNegative(SCIP *scip, SCIP_Real val)
SCIP_RETCODE SCIPreleaseVar(SCIP *scip, SCIP_VAR **var)
Definition: scip_var.c:1248
SCIP_RETCODE SCIPcreateVar(SCIP *scip, SCIP_VAR **var, const char *name, SCIP_Real lb, SCIP_Real ub, SCIP_Real obj, SCIP_VARTYPE vartype, SCIP_Bool initial, SCIP_Bool removable, SCIP_DECL_VARDELORIG((*vardelorig)), SCIP_DECL_VARTRANS((*vartrans)), SCIP_DECL_VARDELTRANS((*vardeltrans)), SCIP_DECL_VARCOPY((*varcopy)), SCIP_VARDATA *vardata)
Definition: scip_var.c:114
SCIP_RETCODE SCIPgetTransformedVar(SCIP *scip, SCIP_VAR *var, SCIP_VAR **transvar)
Definition: scip_var.c:1439
int SCIPsnprintf(char *t, int len, const char *s,...)
Definition: misc.c:10880
Definition: pqueue.h:38
real eps
class for priority queues
SCIP_DECL_PRICERINIT(ObjPricerVRP::scip_init)
Definition: pricer_vrp.cpp:83
SCIP_DECL_PRICERREDCOST(ObjPricerVRP::scip_redcost)
Definition: pricer_vrp.cpp:225
SCIP_DECL_PRICERFARKAS(ObjPricerVRP::scip_farkas)
Definition: pricer_vrp.cpp:246
VRP pricer plugin.
#define SCIPdebugMessage
Definition: pub_message.h:96
@ SCIP_SUCCESS
Definition: type_result.h:58
@ SCIP_OKAY
Definition: type_retcode.h:42
enum SCIP_Retcode SCIP_RETCODE
Definition: type_retcode.h:63
@ SCIP_VARTYPE_CONTINUOUS
Definition: type_var.h:71