0

I am Using the SCIP interactive shell, to solve a problem from an LP file, the solver is running since hours with no improvements in optimality GAP, normally this MIP problem is solved by free Solvers in under one second.

Here is part of console log it's same for hours

 7564s|682700 |669166 | 20689k|  30.3 |4674M| 241 | 110 |1276 |1664 |1276 |1395 |   1 |  31k|  20k| 2.876849e+05 | 2.342078e+05 |  22.83%
 7565s|682800 |669266 | 20693k|  30.3 |4675M| 241 |  61 |1276 |1670 |1276 |1395 |   1 |  31k|  20k| 2.876849e+05 | 2.342078e+05 |  22.83%
 7565s|682900 |669354 | 20699k|  30.3 |4675M| 241 | 129 |1276 |1687 |1276 |1395 |   1 |  31k|  20k| 2.876849e+05 | 2.342078e+05 |  22.83%
 7566s|683000 |669454 | 20701k|  30.3 |4675M| 241 | 102 |1276 |1677 |1276 |1395 |   1 |  31k|  20k| 2.876849e+05 | 2.342078e+05 |  22.83%
 7567s|683100 |669554 | 20706k|  30.3 |4675M| 241 | 149 |1276 |1677 |1276 |1395 |   1 |  31k|  20k| 2.876849e+05 | 2.342078e+05 |  22.83%
 7567s|683200 |669652 | 20709k|  30.3 |4676M| 241 | 106 |1276 |1666 |1276 |1395 |   1 |  31k|  20k| 2.876849e+05 | 2.342078e+05 |  22.83%
 7568s|683300 |669746 | 20713k|  30.3 |4676M| 241 | 112 |1276 |1660 |1276 |1395 |   1 |  31k|  20k| 2.876849e+05 | 2.342078e+05 |  22.83%
 7569s|683400 |669834 | 20716k|  30.3 |4679M| 241 | 139 |1276 |1656 |1276 |1395 |   1 |  31k|  20k| 2.876849e+05 | 2.342078e+05 |  22.83%
 7569s|683500 |669934 | 20721k|  30.3 |4680M| 241 |  93 |1276 |1649 |1276 |1395 |   1 |  31k|  20k| 2.876849e+05 | 2.342078e+05 |  22.83%
 time | node  | left  |LP iter|LP it/n| mem |mdpt |frac |vars |cons |cols |rows |cuts |confs|strbr|  dualbound   | primalbound  |  gap   
 7570s|683600 |670034 | 20724k|  30.3 |4680M| 241 |  92 |1276 |1646 |1276 |1395 |   1 |  31k|  20k| 2.876849e+05 | 2.342078e+05 |  22.83%
 7571s|683700 |670134 | 20727k|  30.3 |4680M| 241 |  78 |1276 |1643 |1276 |1395 |   1 |  31k|  20k| 2.876848e+05 | 2.342078e+05 |  22.83%
 7571s|683800 |670234 | 20730k|  30.3 |4681M| 241 | 141 |1276 |1649 |1276 |1395 |   1 |  31k|  20k| 2.876848e+05 | 2.342078e+05 |  22.83%
 7572s|683900 |670318 | 20733k|  30.3 |4682M| 241 |   - |1276 |1642 |1276

Is there is some kind termination parameter that I am missing, apart form time limit and optimality gap as they provide the non-global solution

Manish
  • 117
  • 1
  • 2
  • 12
  • I don't fully understand what you are asking for, but you can have a look at other termination limits by typing `set limits` in the interactive shell. – mueldgog Aug 27 '19 at 14:42
  • I am asking, that i need to provide scip some way to stop, as MIP GAP is not decreasing. What are those ways – Manish Aug 27 '19 at 15:00
  • Which other free solvers have you tried? Feel free to write to the SCIP mailing list and send your model. – Jakob Aug 27 '19 at 15:07
  • Coin or cbc and glpk – Manish Aug 27 '19 at 15:24
  • Try to perturb the problem a little bit.Also I have seen LP files being interpreted somewhat differently between solvers. An MPS file is safer when using different solvers. – Erwin Kalvelagen Aug 29 '19 at 15:31
  • Hi! I have the same problem while solving some instances of my model. Did you find a way to stop the solver when this happens? – orpanter Sep 06 '22 at 15:20
  • I'm having the problem as well..@orpanter @Manish Did you get any luck? – Steven01123581321 Dec 03 '22 at 06:53

1 Answers1

1

If you want to have full control over the solving process, then you could use an event handler to check the progress in the dual and primal bound by yourself. Here you find an example that shows how to add an event handler to SCIP. In your case, I would suggest catching the SCIP_EVENTTYPE_NODESOLVED event, which is thrown after processing each node of the branch-and-bound tree.

mueldgog
  • 383
  • 1
  • 7