[Scip] Segmentation Fault

Stefan Vigerske stefan at math.hu-berlin.de
Thu Nov 19 11:16:58 MET 2009


Hi,

there is a bug in the SCIP/CLP interface.
The patch is attached.

Stefan

Am 19.11.2009 10:04, schrieb Julio Rojas:
> Dear all, I have been using SCIP-CLP to solve some BIP problems of my PhD
> work. One of the problems has the generic form:
> 
> Min Sum(Xi)
> s.t.
> Xi+Xk <= 1, if Aik=1 and i<k, for i,k in {1,...,n}
> Xi-Xk = 0, if {Xi,Xk} in Pj for all j in {1,...,g}, g<n
> Sum(Xi)>=floor(n/2)+1
> Xi in {0,1}
> 
> I know that for some setups this problem is unfeasible, but sometimes I get
> segmentation fault errors like in the attached example.
> 
> I use the standalone solver. After loading it the message presented is the
> following:
> 
> original problem has 75 variables (0 bin, 75 int, 0 impl, 0 cont) and 846
> constraints
> 
> After presolve, the message is:
> 
> presolving:
> (round 1) 69 del vars, 690 del conss, 0 chg bounds, 0 chg sides, 0 chg
> coeffs, 0 upgd conss, 1046 impls, 0 clqs
> (round 2) 69 del vars, 842 del conss, 0 chg bounds, 0 chg sides, 0 chg
> coeffs, 0 upgd conss, 1046 impls, 0 clqs
> (round 3) 69 del vars, 842 del conss, 0 chg bounds, 0 chg sides, 0 chg
> coeffs, 3 upgd conss, 1046 impls, 0 clqs
> (round 4) 69 del vars, 842 del conss, 0 chg bounds, 0 chg sides, 0 chg
> coeffs, 4 upgd conss, 1046 impls, 0 clqs
> (round 5) 70 del vars, 842 del conss, 0 chg bounds, 2 chg sides, 4 chg
> coeffs, 4 upgd conss, 1056 impls, 0 clqs
> (round 6) 71 del vars, 843 del conss, 0 chg bounds, 3 chg sides, 6 chg
> coeffs, 4 upgd conss, 1060 impls, 0 clqs
> (round 7) 71 del vars, 846 del conss, 1 chg bounds, 5 chg sides, 8 chg
> coeffs, 4 upgd conss, 1062 impls, 0 clqs
> (round 8) 72 del vars, 846 del conss, 1 chg bounds, 5 chg sides, 8 chg
> coeffs, 4 upgd conss, 1062 impls, 0 clqs
> presolving (9 rounds):
>  72 deleted vars, 846 deleted constraints, 1 tightened bounds, 0 added
> holes, 5 changed sides, 8 changed coefficients
>  1062 implications, 0 cliques
> presolved problem has 3 variables (3 bin, 0 int, 0 impl, 0 cont) and 3
> constraints
>       1 constraints of type <knapsack>
>       2 constraints of type <logicor>
> Presolving Time: 0.01
> 
> So, as you can see, the problem should be unfeasible, but when I run the
> solving routine I get the following message:
> 
>  time | node  | left  |LP iter| mem |mdpt |frac |vars |cons |ccons|cols
> |rows |cuts |confs|strbr|  dualbound   | primalbound  |  gap
> t 0.0s|     1 |     0 |     0 | 449k|   0 |   - |   3 |   3 |   3 |   0 |
> 0 |   0 |   0 |   0 |      --      | 1.192662e+03 |    Inf
> t 0.0s|     1 |     0 |     0 | 449k|   0 |   - |   3 |   3 |   3 |   0 |
> 0 |   0 |   0 |   0 |      --      | 1.170286e+03 |    Inf
> Segmentation fault
> 
> Any idea on what's happening? Is there a way to avoid this problem? I'm
> running the standalone solver from an R script, so I don't know if by using
> C I can catch this problem. I don't program in C. Can I trust that when this
> happens the solution is unfeasible? I can cope with this problem if it is
> presented when the solution is unfeasible and some set of features exist in
> the model.
> 
> Thank you very much for your help.
> -------------------------------------------------
> Julio Rojas
> jcredberry at gmail.com
> 
> 
> 
> 
> _______________________________________________
> Scip mailing list
> Scip at zib.de
> http://listserv.zib.de/mailman/listinfo/scip

-------------- next part --------------
A non-text attachment was scrubbed...
Name: scip.patch
Type: text/x-patch
Size: 8272 bytes
Desc: not available
Url : http://listserv.zib.de/mailman/private/scip/attachments/20091119/0a8d5c58/scip.bin


More information about the Scip mailing list