forked from coin-or/Clp
-
Notifications
You must be signed in to change notification settings - Fork 2
/
Copy pathAbcSimplexDual.cpp
6079 lines (6002 loc) · 219 KB
/
AbcSimplexDual.cpp
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
/* $Id$ */
// Copyright (C) 2002, International Business Machines
// Corporation and others, Copyright (C) 2012, FasterCoin. All Rights Reserved.
// This code is licensed under the terms of the Eclipse Public License (EPL).
/* Notes on implementation of dual simplex algorithm.
When dual feasible:
If primal feasible, we are optimal. Otherwise choose an infeasible
basic variable to leave basis (normally going to nearest bound) (B). We
now need to find an incoming variable which will leave problem
dual feasible so we get the row of the tableau corresponding to
the basic variable (with the correct sign depending if basic variable
above or below feasibility region - as that affects whether reduced
cost on outgoing variable has to be positive or negative).
We now perform a ratio test to determine which incoming variable will
preserve dual feasibility (C). If no variable found then problem
is infeasible (in primal sense). If there is a variable, we then
perform pivot and repeat. Trivial?
-------------------------------------------
A) How do we get dual feasible? If all variables have bounds then
it is trivial to get feasible by putting non-basic variables to
correct bounds. OSL did not have a phase 1/phase 2 approach but
instead effectively put fake bounds on variables and this is the
approach here, although I had hoped to make it cleaner.
If there is a weight of X on getting dual feasible:
Non-basic variables with negative reduced costs are put to
lesser of their upper bound and their lower bound + X.
Similarly, mutatis mutandis, for positive reduced costs.
Free variables should normally be in basis, otherwise I have
coding which may be able to come out (and may not be correct).
In OSL, this weight was changed heuristically, here at present
it is only increased if problem looks finished. If problem is
feasible I check for unboundedness. If not unbounded we
could play with going into primal. As long as weights increase
any algorithm would be finite.
B) Which outgoing variable to choose is a virtual base class.
For difficult problems steepest edge is preferred while for
very easy (large) problems we will need partial scan.
C) Sounds easy, but this is hardest part of algorithm.
1) Instead of stopping at first choice, we may be able
to flip that variable to other bound and if objective
still improving choose again. These mini iterations can
increase speed by orders of magnitude but we may need to
go to more of a bucket choice of variable rather than looking
at them one by one (for speed).
2) Accuracy. Reduced costs may be of wrong sign but less than
tolerance. Pivoting on these makes objective go backwards.
OSL modified cost so a zero move was made, Gill et al
(in primal analogue) modified so a strictly positive move was
made. It is not quite as neat in dual but that is what we
try and do. The two problems are that re-factorizations can
change reduced costs above and below tolerances and that when
finished we need to reset costs and try again.
3) Degeneracy. Gill et al helps but may not be enough. We
may need more. Also it can improve speed a lot if we perturb
the costs significantly.
References:
Forrest and Goldfarb, Steepest-edge simplex algorithms for
linear programming - Mathematical Programming 1992
Forrest and Tomlin, Implementing the simplex method for
the Optimization Subroutine Library - IBM Systems Journal 1992
Gill, Murray, Saunders, Wright A Practical Anti-Cycling
Procedure for Linear and Nonlinear Programming SOL report 1988
TODO:
a) Better recovery procedures. At present I never check on forward
progress. There is checkpoint/restart with reducing
re-factorization frequency, but this is only on singular
factorizations.
b) Fast methods for large easy problems (and also the option for
the code to automatically choose which method).
c) We need to be able to stop in various ways for OSI - this
is fairly easy.
*/
#include "CoinPragma.hpp"
#include <math.h>
#include "CoinHelperFunctions.hpp"
#include "CoinAbcHelperFunctions.hpp"
#include "AbcSimplexDual.hpp"
#include "ClpEventHandler.hpp"
#include "AbcSimplexFactorization.hpp"
#include "AbcNonLinearCost.hpp"
#include "CoinPackedMatrix.hpp"
#include "CoinIndexedVector.hpp"
#include "CoinFloatEqual.hpp"
#include "AbcDualRowDantzig.hpp"
#include "AbcDualRowSteepest.hpp"
#include "ClpMessage.hpp"
#include "ClpLinearObjective.hpp"
class ClpSimplex;
#include <cfloat>
#include <cassert>
#include <string>
#include <stdio.h>
#include <iostream>
//#define CLP_DEBUG 1
#ifdef NDEBUG
#define NDEBUG_CLP
#endif
#ifndef CLP_INVESTIGATE
#define NDEBUG_CLP
#endif
// dual
/* *** Method
This is a vanilla version of dual simplex.
It tries to be a single phase approach with a weight of 1.0 being
given to getting optimal and a weight of dualBound_ being
given to getting dual feasible. In this version I have used the
idea that this weight can be thought of as a fake bound. If the
distance between the lower and upper bounds on a variable is less
than the feasibility weight then we are always better off flipping
to other bound to make dual feasible. If the distance is greater
then we make up a fake bound dualBound_ away from one bound.
If we end up optimal or primal infeasible, we check to see if
bounds okay. If so we have finished, if not we increase dualBound_
and continue (after checking if unbounded). I am undecided about
free variables - there is coding but I am not sure about it. At
present I put them in basis anyway.
The code is designed to take advantage of sparsity so arrays are
seldom zeroed out from scratch or gone over in their entirety.
The only exception is a full scan to find outgoing variable. This
will be changed to keep an updated list of infeasibilities (or squares
if steepest edge). Also on easy problems we don't need full scan - just
pick first reasonable.
One problem is how to tackle degeneracy and accuracy. At present
I am using the modification of costs which I put in OSL and which was
extended by Gill et al. I am still not sure of the exact details.
The flow of dual is three while loops as follows:
while (not finished) {
while (not clean solution) {
Factorize and/or clean up solution by flipping variables so
dual feasible. If looks finished check fake dual bounds.
Repeat until status is iterating (-1) or finished (0,1,2)
}
while (status==-1) {
Iterate until no pivot in or out or time to re-factorize.
Flow is:
choose pivot row (outgoing variable). if none then
we are primal feasible so looks as if done but we need to
break and check bounds etc.
Get pivot row in tableau
Choose incoming column. If we don't find one then we look
primal infeasible so break and check bounds etc. (Also the
pivot tolerance is larger after any iterations so that may be
reason)
If we do find incoming column, we may have to adjust costs to
keep going forwards (anti-degeneracy). Check pivot will be stable
and if unstable throw away iteration (we will need to implement
flagging of basic variables sometime) and break to re-factorize.
If minor error re-factorize after iteration.
Update everything (this may involve flipping variables to stay
dual feasible.
}
}
At present we never check we are going forwards. I overdid that in
OSL so will try and make a last resort.
Needs partial scan pivot out option.
Needs dantzig, uninitialized and full steepest edge options (can still
use partial scan)
May need other anti-degeneracy measures, especially if we try and use
loose tolerances as a way to solve in fewer iterations.
I like idea of dynamic scaling. This gives opportunity to decouple
different implications of scaling for accuracy, iteration count and
feasibility tolerance.
*/
#define CLEAN_FIXED 0
// Startup part of dual (may be extended to other algorithms)
// To force to follow another run put logfile name here and define
//#define FORCE_FOLLOW
#ifdef FORCE_FOLLOW
static FILE *fpFollow = NULL;
static const char *forceFile = NULL;
static int force_in = -1;
static int force_out = -1;
static int force_way_in = -1;
static int force_way_out = -1;
static int force_iterations = 0;
int force_argc = 0;
const char **force_argv = NULL;
#endif
void AbcSimplexDual::startupSolve()
{
#ifdef FORCE_FOLLOW
int ifld;
for (ifld = 1; ifld < force_argc; ifld++) {
if (strstr(argv[ifld], ".log")) {
forceFile = argv[ifld];
break;
}
}
if (!fpFollow && strlen(forceFile)) {
fpFollow = fopen(forceFile, "r");
assert(fpFollow);
}
if (fpFollow) {
int numberRead = atoi(argv[ifld + 1]);
force_iterations = atoi(argv[ifld + 1]);
printf("skipping %d iterations and then doing %d\n", numberRead, force_iterations);
for (int iteration = 0; iteration < numberRead; iteration++) {
// read to next Clp0102
char temp[300];
while (fgets(temp, 250, fpFollow)) {
if (!strncmp(temp, "dirOut", 6))
break;
}
sscanf(temp + 7, "%d%*s%d%*s%*s%d",
&force_way_out, &force_way_in);
fgets(temp, 250, fpFollow);
char cin, cout;
int whichIteration;
sscanf(temp, "%d%*f%*s%*c%c%d%*s%*c%c%d",
&whichIteration, &cin, &force_in, &cout, &force_out);
assert(whichIterations == iteration + 1);
if (cin == 'C')
force_in += numberColumns_;
if (cout == 'C')
force_out += numberColumns_;
printf("Iteration %d will force %d out (way %d) and %d in (way %d)\n",
whichIteration, force_out, force_way_out, force_in, force_way_in);
assert(getInternalStatus(force_out) == basic);
assert(getInternalStatus(force_in) != basic);
setInternalStatus(force_in, basic);
if (force_way_out == -1)
setInternalStatus(force_out, atUpperBound);
else
setInternalStatus(force_out, atLowerBound);
}
// clean up
int numberBasic = 0;
for (int i = 0; i < numberTotal_; i++) {
if (getInternalStatus(i) == basic) {
abcPivotVariable_[numberBasic++] = i;
} else if (getInternalStatus(i) == atLowerBound) {
if (abcLower_[i] < -1.0e30) {
abcLower_[i] = abcUpper_[i] - dualBound_;
setFakeLower(i);
}
} else if (getInternalStatus(i) == atUpperBound) {
if (abcUpper_[i] > 1.0e30) {
abcUpper_[i] = abcLower_[i] + dualBound_;
setFakeUpper(i);
}
}
}
assert(numberBasic == numberRows_);
}
#endif
// initialize - no values pass and algorithm_ is -1
// put in standard form (and make row copy)
// create modifiable copies of model rim and do optional scaling
// If problem looks okay
// Do initial factorization
numberFake_ = 0; // Number of variables at fake bounds
numberChanged_ = 0; // Number of variables with changed costs
/// Initial coding here
statusOfProblemInDual(0);
}
void AbcSimplexDual::finishSolve()
{
assert(problemStatus_ || !sumPrimalInfeasibilities_);
}
void AbcSimplexDual::gutsOfDual()
{
double largestPrimalError = 0.0;
double largestDualError = 0.0;
lastCleaned_ = 0; // last time objective or bounds cleaned up
numberDisasters_ = 0;
// This says whether to restore things etc
//startupSolve();
// startup will have factorized so can skip
// Start check for cycles
abcProgress_.startCheck();
// Say change made on first iteration
changeMade_ = 1;
// Say last objective infinite
//lastObjectiveValue_=-COIN_DBL_MAX;
if ((stateOfProblem_ & VALUES_PASS) != 0) {
// modify costs so nonbasic dual feasible with fake
for (int iSequence = 0; iSequence < numberTotal_; iSequence++) {
switch (getInternalStatus(iSequence)) {
case basic:
case AbcSimplex::isFixed:
break;
case isFree:
case superBasic:
//abcCost_[iSequence]-=djSaved_[iSequence];
//abcDj_[iSequence]-=djSaved_[iSequence];
djSaved_[iSequence] = 0.0;
break;
case atLowerBound:
if (djSaved_[iSequence] < 0.0) {
//abcCost_[iSequence]-=djSaved_[iSequence];
//abcDj_[iSequence]-=djSaved_[iSequence];
djSaved_[iSequence] = 0.0;
}
break;
case atUpperBound:
if (djSaved_[iSequence] > 0.0) {
//abcCost_[iSequence]-=djSaved_[iSequence];
//abcDj_[iSequence]-=djSaved_[iSequence];
djSaved_[iSequence] = 0.0;
}
break;
}
}
}
progressFlag_ = 0;
/*
Status of problem:
0 - optimal
1 - infeasible
2 - unbounded
-1 - iterating
-2 - factorization wanted
-3 - redo checking without factorization
-4 - looks infeasible
*/
int factorizationType = 1;
double pivotTolerance = abcFactorization_->pivotTolerance();
while (problemStatus_ < 0) {
// If getting nowhere - return
double limit = numberRows_ + numberColumns_;
limit = 10000.0 + 2000.0 * limit;
#if ABC_NORMAL_DEBUG > 0
if (numberIterations_ > limit) {
printf("Set flag so ClpSimplexDual done\n");
abort();
}
#endif
bool disaster = false;
if (disasterArea_ && disasterArea_->check()) {
disasterArea_->saveInfo();
disaster = true;
}
if (numberIterations_ > baseIteration_) {
// may factorize, checks if problem finished
statusOfProblemInDual(factorizationType);
factorizationType = 1;
largestPrimalError = CoinMax(largestPrimalError, largestPrimalError_);
largestDualError = CoinMax(largestDualError, largestDualError_);
if (disaster)
problemStatus_ = 3;
}
int saveNumberIterations = numberIterations_;
// exit if victory declared
if (problemStatus_ >= 0)
break;
// test for maximum iterations
if (hitMaximumIterations()) {
problemStatus_ = 3;
break;
}
// Check event
int eventStatus = eventHandler_->event(ClpEventHandler::endOfFactorization);
if (eventStatus >= 0) {
problemStatus_ = 5;
secondaryStatus_ = ClpEventHandler::endOfFactorization;
break;
}
if (!numberPrimalInfeasibilities_ && problemStatus_ < 0) {
#if ABC_NORMAL_DEBUG > 0
printf("Primal feasible - sum dual %g - go to primal\n", sumDualInfeasibilities_);
#endif
problemStatus_ = 10;
break;
}
// Do iterations ? loop round
#if PARTITION_ROW_COPY == 1
stateOfProblem_ |= NEED_BASIS_SORT;
#endif
while (true) {
#if PARTITION_ROW_COPY == 1
if ((stateOfProblem_ & NEED_BASIS_SORT) != 0) {
int iVector = getAvailableArray();
abcMatrix_->sortUseful(usefulArray_[iVector]);
setAvailableArray(iVector);
stateOfProblem_ &= ~NEED_BASIS_SORT;
}
#endif
problemStatus_ = -1;
if ((stateOfProblem_ & VALUES_PASS) != 0) {
assert(djSaved_ - abcDj_ == numberTotal_);
abcDj_ = djSaved_;
djSaved_ = NULL;
}
#if ABC_PARALLEL
if (parallelMode_ == 0) {
#endif
whileIteratingSerial();
#if ABC_PARALLEL
} else {
#if ABC_PARALLEL == 1
whileIteratingThread();
#else
whileIteratingCilk();
#endif
}
#endif
if ((stateOfProblem_ & VALUES_PASS) != 0) {
djSaved_ = abcDj_;
abcDj_ = djSaved_ - numberTotal_;
}
if (pivotRow_ != -100) {
if (numberIterations_ > saveNumberIterations || problemStatus_ >= 0) {
if (problemStatus_ == 10 && abcFactorization_->pivots() && numberTimesOptimal_ < 3) {
factorizationType = 5;
problemStatus_ = -4;
if (!numberPrimalInfeasibilities_)
factorizationType = 9;
} else if (problemStatus_ == -2) {
factorizationType = 5;
}
break;
} else if (abcFactorization_->pivots() || problemStatus_ == -6 || pivotTolerance < abcFactorization_->pivotTolerance()) {
factorizationType = 5;
if (pivotTolerance < abcFactorization_->pivotTolerance()) {
pivotTolerance = abcFactorization_->pivotTolerance();
if (!abcFactorization_->pivots())
abcFactorization_->minimumPivotTolerance(CoinMin(0.25, 1.05 * abcFactorization_->minimumPivotTolerance()));
}
break;
} else {
bool tryFake = numberAtFakeBound() > 0 && currentDualBound_ < 1.0e17;
if (problemStatus_ == -5 || tryFake) {
if (numberTimesOptimal_ > 10)
problemStatus_ = 4;
numberTimesOptimal_++;
int numberChanged = -1;
if (numberFlagged_) {
numberFlagged_ = 0;
for (int iRow = 0; iRow < numberRows_; iRow++) {
int iPivot = abcPivotVariable_[iRow];
if (flagged(iPivot)) {
clearFlagged(iPivot);
}
}
numberChanged = 0;
} else if (tryFake) {
double dummyChange;
numberChanged = changeBounds(0, dummyChange);
if (numberChanged > 0) {
handler_->message(CLP_DUAL_ORIGINAL, messages_)
<< CoinMessageEol;
bounceTolerances(3 /*(numberChanged>0) ? 3 : 1*/);
// temp
#if ABC_NORMAL_DEBUG > 0
//printf("should not need to break out of loop\n");
#endif
abcProgress_.reset();
break;
} else if (numberChanged < -1) {
// some flipped
bounceTolerances(3);
abcProgress_.reset();
}
} else if (numberChanged < 0) {
// what now
printf("No iterations since last inverts A - nothing done - think\n");
abort();
}
} else if (problemStatus_ != -4) {
// what now
printf("No iterations since last inverts B - nothing done - think\n");
abort();
}
}
} else {
// end of values pass
// we need to tell statusOf.. to restore costs etc
// and take off flag
bounceTolerances(2);
assert(!solution_);
delete[] solution_;
solution_ = NULL;
gutsOfSolution(3);
makeNonFreeVariablesDualFeasible();
stateOfProblem_ &= ~VALUES_PASS;
problemStatus_ = -2;
factorizationType = 5;
break;
}
}
if (problemStatus_ == 1 && (progressFlag_ & 8) != 0 && fabs(rawObjectiveValue_) > 1.0e10) {
#if ABC_NORMAL_DEBUG > 0
printf("fix looks infeasible when has been\n"); // goto primal
#endif
problemStatus_ = 10; // infeasible - but has looked feasible
}
if (!problemStatus_ && abcFactorization_->pivots()) {
gutsOfSolution(1); // need to compute duals
//computeInternalObjectiveValue();
checkCutoff(true);
}
}
largestPrimalError_ = largestPrimalError;
largestDualError_ = largestDualError;
}
/// The duals are updated by the given arrays.
static void updateDualsInDualBit2(CoinPartitionedVector &row,
double *COIN_RESTRICT djs,
double theta, int iBlock)
{
int offset = row.startPartition(iBlock);
double *COIN_RESTRICT work = row.denseVector() + offset;
const int *COIN_RESTRICT which = row.getIndices() + offset;
int number = row.getNumElements(iBlock);
for (int i = 0; i < number; i++) {
int iSequence = which[i];
double alphaI = work[i];
work[i] = 0.0;
double value = djs[iSequence] - theta * alphaI;
djs[iSequence] = value;
}
row.setNumElementsPartition(iBlock, 0);
}
static void updateDualsInDualBit(CoinPartitionedVector &row,
double *djs,
double theta, int numberBlocks)
{
for (int iBlock = 1; iBlock < numberBlocks; iBlock++) {
cilk_spawn updateDualsInDualBit2(row, djs, theta, iBlock);
}
updateDualsInDualBit2(row, djs, theta, 0);
cilk_sync;
}
/// The duals are updated by the given arrays.
void AbcSimplexDual::updateDualsInDual()
{
//dual->usefulArray(3)->compact();
CoinPartitionedVector &array = usefulArray_[arrayForTableauRow_];
//array.compact();
int numberBlocks = array.getNumPartitions();
int number = array.getNumElements();
if (numberBlocks) {
if (number > 100) {
updateDualsInDualBit(array, abcDj_, theta_, numberBlocks);
} else {
for (int iBlock = 0; iBlock < numberBlocks; iBlock++) {
updateDualsInDualBit2(array, abcDj_, theta_, iBlock);
}
}
array.compact();
} else {
double *COIN_RESTRICT work = array.denseVector();
const int *COIN_RESTRICT which = array.getIndices();
double *COIN_RESTRICT reducedCost = abcDj_;
#pragma cilk grainsize = 128
cilk_for(int i = 0; i < number; i++)
{
//for (int i = 0; i < number; i++) {
int iSequence = which[i];
double alphaI = work[i];
work[i] = 0.0;
double value = reducedCost[iSequence] - theta_ * alphaI;
reducedCost[iSequence] = value;
}
array.setNumElements(0);
}
}
int AbcSimplexDual::nextSuperBasic()
{
if (firstFree_ >= 0) {
// make sure valid
int returnValue = firstFree_;
int iColumn = firstFree_ + 1;
#ifdef PAN
if (fakeSuperBasic(firstFree_) < 0) {
// somehow cleaned up
returnValue = -1;
for (; iColumn < numberTotal_; iColumn++) {
if (getInternalStatus(iColumn) == isFree || getInternalStatus(iColumn) == superBasic) {
if (fakeSuperBasic(iColumn) >= 0) {
if (fabs(abcDj_[iColumn]) > 1.0e2 * dualTolerance_)
break;
}
}
}
if (iColumn < numberTotal_) {
returnValue = iColumn;
iColumn++;
}
}
#endif
for (; iColumn < numberTotal_; iColumn++) {
if (getInternalStatus(iColumn) == isFree || getInternalStatus(iColumn) == superBasic) {
#ifdef PAN
if (fakeSuperBasic(iColumn) >= 0) {
#endif
if (fabs(abcDj_[iColumn]) > dualTolerance_)
break;
#ifdef PAN
}
#endif
}
}
firstFree_ = iColumn;
if (firstFree_ == numberTotal_)
firstFree_ = -1;
return returnValue;
} else {
return -1;
}
}
/*
Chooses dual pivot row
For easy problems we can just choose one of the first rows we look at
*/
void AbcSimplexDual::dualPivotRow()
{
//int lastPivotRow=pivotRow_;
pivotRow_ = -1;
const double *COIN_RESTRICT lowerBasic = lowerBasic_;
const double *COIN_RESTRICT upperBasic = upperBasic_;
const double *COIN_RESTRICT solutionBasic = solutionBasic_;
const int *COIN_RESTRICT pivotVariable = abcPivotVariable_;
freeSequenceIn_ = -1;
double worstValuesPass = 0.0;
if ((stateOfProblem_ & VALUES_PASS) != 0) {
for (int iSequence = 0; iSequence < numberTotal_; iSequence++) {
switch (getInternalStatus(iSequence)) {
case basic:
#if ABC_NORMAL_DEBUG > 0
if (fabs(abcDj_[iSequence]) > 1.0e-6)
printf("valuespass basic dj %d status %d dj %g\n", iSequence,
getInternalStatus(iSequence), abcDj_[iSequence]);
#endif
break;
case AbcSimplex::isFixed:
break;
case isFree:
case superBasic:
#if ABC_NORMAL_DEBUG > 0
if (fabs(abcDj_[iSequence]) > 1.0e-6)
printf("Bad valuespass dj %d status %d dj %g\n", iSequence,
getInternalStatus(iSequence), abcDj_[iSequence]);
#endif
break;
case atLowerBound:
#if ABC_NORMAL_DEBUG > 0
if (abcDj_[iSequence] < -1.0e-6)
printf("Bad valuespass dj %d status %d dj %g\n", iSequence,
getInternalStatus(iSequence), abcDj_[iSequence]);
#endif
break;
case atUpperBound:
#if ABC_NORMAL_DEBUG > 0
if (abcDj_[iSequence] > 1.0e-6)
printf("Bad valuespass dj %d status %d dj %g\n", iSequence,
getInternalStatus(iSequence), abcDj_[iSequence]);
#endif
break;
}
}
// get worst basic dual - later do faster using steepest infeasibility array
// does not have to be worst - just bad
int iWorst = -1;
double worst = dualTolerance_;
for (int i = 0; i < numberRows_; i++) {
int iSequence = abcPivotVariable_[i];
if (fabs(abcDj_[iSequence]) > worst) {
iWorst = i;
worst = fabs(abcDj_[iSequence]);
}
}
if (iWorst >= 0) {
pivotRow_ = iWorst;
worstValuesPass = abcDj_[abcPivotVariable_[iWorst]];
} else {
// factorize and clean up costs
pivotRow_ = -100;
return;
}
}
if (false && numberFreeNonBasic_ > abcFactorization_->maximumPivots()) {
lastFirstFree_ = -1;
firstFree_ = -1;
}
if (firstFree_ >= 0) {
// free
int nextFree = nextSuperBasic();
if (nextFree >= 0) {
freeSequenceIn_ = nextFree;
// **** should skip price (and maybe weights update)
// unpack vector and find a good pivot
unpack(usefulArray_[arrayForBtran_], nextFree);
abcFactorization_->updateColumn(usefulArray_[arrayForBtran_]);
double *COIN_RESTRICT work = usefulArray_[arrayForBtran_].denseVector();
int number = usefulArray_[arrayForBtran_].getNumElements();
int *COIN_RESTRICT which = usefulArray_[arrayForBtran_].getIndices();
double bestFeasibleAlpha = 0.0;
int bestFeasibleRow = -1;
double bestInfeasibleAlpha = 0.0;
int bestInfeasibleRow = -1;
// Go for big infeasibilities unless big changes
double maxInfeas = (fabs(dualOut_) > 1.0e3) ? 0.001 : 1.0e6;
for (int i = 0; i < number; i++) {
int iRow = which[i];
double alpha = fabs(work[iRow]);
if (alpha > 1.0e-3) {
double value = solutionBasic[iRow];
double lower = lowerBasic[iRow];
double upper = upperBasic[iRow];
double infeasibility = 0.0;
if (value > upper)
infeasibility = value - upper;
else if (value < lower)
infeasibility = lower - value;
infeasibility = CoinMin(maxInfeas, infeasibility);
if (infeasibility * alpha > bestInfeasibleAlpha && alpha > 1.0e-1) {
if (!flagged(pivotVariable[iRow])) {
bestInfeasibleAlpha = infeasibility * alpha;
bestInfeasibleRow = iRow;
}
}
if (alpha > bestFeasibleAlpha && (lower > -1.0e20 || upper < 1.0e20)) {
bestFeasibleAlpha = alpha;
bestFeasibleRow = iRow;
}
}
}
if (bestInfeasibleRow >= 0)
pivotRow_ = bestInfeasibleRow;
else if (bestFeasibleAlpha > 1.0e-2)
pivotRow_ = bestFeasibleRow;
usefulArray_[arrayForBtran_].clear();
if (pivotRow_ < 0) {
// no good
freeSequenceIn_ = -1;
if (!abcFactorization_->pivots()) {
lastFirstFree_ = -1;
firstFree_ = -1;
}
}
}
}
if (pivotRow_ < 0) {
// put back so can test
//pivotRow_=lastPivotRow;
// get pivot row using whichever method it is
pivotRow_ = abcDualRowPivot_->pivotRow();
}
if (pivotRow_ >= 0) {
sequenceOut_ = pivotVariable[pivotRow_];
valueOut_ = solutionBasic[pivotRow_];
lowerOut_ = lowerBasic[pivotRow_];
upperOut_ = upperBasic[pivotRow_];
// if we have problems we could try other way and hope we get a
// zero pivot?
if (valueOut_ > upperOut_) {
directionOut_ = -1;
dualOut_ = valueOut_ - upperOut_;
} else if (valueOut_ < lowerOut_) {
directionOut_ = 1;
dualOut_ = lowerOut_ - valueOut_;
} else {
// odd (could be free) - it's feasible - go to nearest
if (valueOut_ - lowerOut_ < upperOut_ - valueOut_) {
directionOut_ = 1;
dualOut_ = lowerOut_ - valueOut_;
} else {
directionOut_ = -1;
dualOut_ = valueOut_ - upperOut_;
}
}
if (worstValuesPass) {
// in values pass so just use sign of dj
// We don't want to go through any barriers so set dualOut low
// free variables will never be here
// need to think about tolerances with inexact solutions
// could try crashing in variables away from bounds and with near zero djs
// pivot to take out bad slack and increase objective
double fakeValueOut = solution_[sequenceOut_];
dualOut_ = 1.0e-10;
if (upperOut_ > lowerOut_ || fabs(fakeValueOut - upperOut_) < 100.0 * primalTolerance_) {
if (abcDj_[sequenceOut_] > 0.0) {
directionOut_ = 1;
} else {
directionOut_ = -1;
}
} else {
// foolishly trust primal solution
if (fakeValueOut > upperOut_) {
directionOut_ = -1;
} else {
directionOut_ = 1;
}
}
#if 1
// make sure plausible but only when fake primals stored
if (directionOut_ < 0 && fabs(fakeValueOut - upperOut_) > dualBound_ + primalTolerance_) {
if (fabs(fakeValueOut - upperOut_) > fabs(fakeValueOut - lowerOut_))
directionOut_ = 1;
} else if (directionOut_ > 0 && fabs(fakeValueOut - lowerOut_) > dualBound_ + primalTolerance_) {
if (fabs(fakeValueOut - upperOut_) < fabs(fakeValueOut - lowerOut_))
directionOut_ = -1;
}
#endif
}
#if ABC_NORMAL_DEBUG > 3
assert(dualOut_ >= 0.0);
#endif
} else {
sequenceOut_ = -1;
}
alpha_ = 0.0;
upperTheta_ = COIN_DBL_MAX;
return;
}
// Checks if any fake bounds active - if so returns number and modifies
// dualBound_ and everything.
// Free variables will be left as free
// Returns number of bounds changed if >=0
// Returns -1 if not initialize and no effect
// Fills in changeVector which can be used to see if unbounded
// initialize is 0 or 3
// and cost of change vector
int AbcSimplexDual::changeBounds(int initialize,
double &changeCost)
{
assert(initialize == 0 || initialize == 1 || initialize == 3 || initialize == 4);
numberFake_ = 0;
double *COIN_RESTRICT abcLower = abcLower_;
double *COIN_RESTRICT abcUpper = abcUpper_;
double *COIN_RESTRICT abcSolution = abcSolution_;
double *COIN_RESTRICT abcCost = abcCost_;
if (!initialize) {
int numberInfeasibilities;
double newBound;
newBound = 5.0 * currentDualBound_;
numberInfeasibilities = 0;
changeCost = 0.0;
// put back original bounds and then check
CoinAbcMemcpy(abcLower, lowerSaved_, numberTotal_);
CoinAbcMemcpy(abcUpper, upperSaved_, numberTotal_);
const double *COIN_RESTRICT dj = abcDj_;
int iSequence;
// bounds will get bigger - just look at ones at bounds
int numberFlipped = 0;
double checkBound = 1.000000000001 * currentDualBound_;
for (iSequence = 0; iSequence < numberTotal_; iSequence++) {
double lowerValue = abcLower[iSequence];
double upperValue = abcUpper[iSequence];
double value = abcSolution[iSequence];
setFakeBound(iSequence, AbcSimplexDual::noFake);
switch (getInternalStatus(iSequence)) {
case basic:
case AbcSimplex::isFixed:
break;
case isFree:
case superBasic:
break;
case atUpperBound:
if (fabs(value - upperValue) > primalTolerance_) {
// can we flip
if (value - lowerValue < checkBound && dj[iSequence] > -currentDualTolerance_) {
abcSolution_[iSequence] = lowerValue;
setInternalStatus(iSequence, atLowerBound);
numberFlipped++;
} else {
numberInfeasibilities++;
}
}
break;
case atLowerBound:
if (fabs(value - lowerValue) > primalTolerance_) {
// can we flip
if (upperValue - value < checkBound && dj[iSequence] < currentDualTolerance_) {
abcSolution_[iSequence] = upperValue;
setInternalStatus(iSequence, atUpperBound);
numberFlipped++;
} else {
numberInfeasibilities++;
}
}
break;
}
}
// If dual infeasible then carry on
if (numberInfeasibilities) {
handler_->message(CLP_DUAL_CHECKB, messages_)
<< newBound
<< CoinMessageEol;
int iSequence;
for (iSequence = 0; iSequence < numberRows_ + numberColumns_; iSequence++) {
double lowerValue = abcLower[iSequence];
double upperValue = abcUpper[iSequence];
double newLowerValue;
double newUpperValue;
Status status = getInternalStatus(iSequence);
if (status == atUpperBound || status == atLowerBound) {
double value = abcSolution[iSequence];
if (value - lowerValue <= upperValue - value) {
newLowerValue = CoinMax(lowerValue, value - 0.666667 * newBound);
newUpperValue = CoinMin(upperValue, newLowerValue + newBound);
} else {
newUpperValue = CoinMin(upperValue, value + 0.666667 * newBound);
newLowerValue = CoinMax(lowerValue, newUpperValue - newBound);
}
abcLower[iSequence] = newLowerValue;
abcUpper[iSequence] = newUpperValue;
if (newLowerValue > lowerValue) {
if (newUpperValue < upperValue) {
setFakeBound(iSequence, AbcSimplexDual::bothFake);
#ifdef CLP_INVESTIGATE
abort(); // No idea what should happen here - I have never got here
#endif
numberFake_++;
} else {
setFakeBound(iSequence, AbcSimplexDual::lowerFake);
assert(abcLower[iSequence] > lowerSaved_[iSequence]);
assert(abcUpper[iSequence] == upperSaved_[iSequence]);
numberFake_++;
}
} else {
if (newUpperValue < upperValue) {
setFakeBound(iSequence, AbcSimplexDual::upperFake);
assert(abcLower[iSequence] == lowerSaved_[iSequence]);
assert(abcUpper[iSequence] < upperSaved_[iSequence]);
numberFake_++;
}
}
if (status == atUpperBound)
abcSolution[iSequence] = newUpperValue;
else
abcSolution[iSequence] = newLowerValue;
double movement = abcSolution[iSequence] - value;
changeCost += movement * abcCost[iSequence];
}
}
currentDualBound_ = newBound;
#if ABC_NORMAL_DEBUG > 0
printf("new dual bound %g\n", currentDualBound_);
#endif
} else {
numberInfeasibilities = -1 - numberFlipped;
}
return numberInfeasibilities;
} else {
int iSequence;
if (initialize == 3) {
for (iSequence = 0; iSequence < numberTotal_; iSequence++) {
#if ABC_NORMAL_DEBUG > 1
int bad = -1;
if (getFakeBound(iSequence) == noFake) {
if (abcLower_[iSequence] != lowerSaved_[iSequence] || abcUpper_[iSequence] != upperSaved_[iSequence])
bad = 0;
} else if (getFakeBound(iSequence) == lowerFake) {
if (abcLower_[iSequence] == lowerSaved_[iSequence] || abcUpper_[iSequence] != upperSaved_[iSequence])
bad = 1;
} else if (getFakeBound(iSequence) == upperFake) {
if (abcLower_[iSequence] != lowerSaved_[iSequence] || abcUpper_[iSequence] == upperSaved_[iSequence])
bad = 2;
} else {