[SCIP] LP numerical troubles.

[James Cussens]

Hi Marc,

Thanks. That got me what I wanted (after adding the line #include "scip/scip_lp.h"  to solve.c). Turns out the troublesome LP is infeasible, which gives me something to go on to work out what is going wrong here.

James

James Cussens
Room MVB 3.26
Dept of Computer Science, University of Bristol
Phone: +44 (0)117 455 8723
https://jcussens.github.io/
Funded PhDs available in Bristol in the following areas: Data Science<http://www.bristol.ac.uk/cdt/compass/>, Interactive AI<http://www.bristol.ac.uk/cdt/interactive-ai/>, Cyber Security<http://www.bristol.ac.uk/cdt/cyber-security/> or Digital Health<http://www.bristol.ac.uk/cdt/digital-health/>.
________________________________
From: Scip <scip-bounces at zib.de> on behalf of Marc Pfetsch <pfetsch at mathematik.tu-darmstadt.de>
Sent: 28 October 2022 16:00
To: scip at zib.de <scip at zib.de>
Subject: Re: [SCIP] LP numerical troubles.



Hi James,

you could add

SCIP_CALL( SCIPwriteLP(set->scip, "debug.lp") );

at solve.c:4624. This will write out the current LP.

Also a good idea is to turn on the parameter "display/lpinfo". This will
show the output of the LP-solver.

Best

Marc




On 28/10/2022 13:55, James Cussens wrote:
> ​Hi all,
>
> In a SCIP project I am working on I am continually running into
> "unresolved numerical troubles in LP". Some example output given below.
> I get this whether using SOPLEX or CPLEX and whether or not I link with
> IPOT. I suspect that I am somehow generating "weird" LPs so I would like
> to inspect the LPs that are causing these problems. What's the best way
> to do this? (Btw, I am using both pricing and cutting in the
> price-and-cut loop.)
>
> I'm using SCIP 8.0.2
>
> James
>
>
> time | node  | left  |LP iter|LP it/n|mem/heur|mdpt |frac |vars |cons
> |cuts |sepa|confs|strbr|  dualbound   | primalbound  |  gap   | compl.
> p 3.7s|     1 |     0 |    14 |     - |shiftand|   0 |   - |9815 | 853 |
>    0 |  0 |   6 |   0 |      --      |-5.916805e+03 |    Inf | unknown
> (node 1) unresolved numerical troubles in LP 5 -- using pseudo solution
> instead (loop 1)
> k 3.8s|     1 |     0 |    16 |     - |roundand|   0 |   - |9815 | 853 |
>    0 |  0 |   6 |   0 |      --      |-2.814319e+03 |    Inf | unknown
>    3.9s|     1 |     2 |    16 |     - |    91M |   0 |   - |9815 | 853
> |   0 |  0 |   6 |   0 |      --      |-2.814319e+03 |    Inf | unknown
>    3.9s|     2 |     3 |    18 |  16.0 |    91M |   1 |   - |9815 | 853
> |   0 |  0 |   6 |   0 |      --      |-2.814319e+03 |    Inf | unknown
>    4.0s|     3 |     4 |    22 |  10.0 |    91M |   2 |   - |9815 | 853
> |   0 |  0 |   6 |   0 |      --      |-2.814319e+03 |    Inf | unknown
>    4.0s|     4 |     5 |    24 |   7.3 |    91M |   3 |   - |9815 | 853
> |   0 |  0 |   6 |   0 |      --      |-2.814319e+03 |    Inf | unknown
>    4.0s|     5 |     6 |    26 |   6.0 |    91M |   4 |   - |9815 | 853
> |   0 |  0 |   6 |   0 |      --      |-2.814319e+03 |    Inf | unknown
>    4.0s|     6 |     7 |    28 |   5.2 |    91M |   5 |   - |9815 | 853
> |   0 |  0 |   6 |   0 |      --      |-2.814319e+03 |    Inf | unknown
> [solve.c:4624] ERROR: (node 7) unresolved numerical troubles in LP 10 --
> aborting
> [solve.c:4984] ERROR: Error <-6> in function call
> [scip_solve.c:2771] ERROR: Error <-6> in function call
> [scipshell.c:101] ERROR: Error <-6> in function call
> [scipshell.c:468] ERROR: Error <-6> in function call
> [cmain.c:104] ERROR: Error <-6> in function call
> SCIP Error (-6): error in LP solver
>
>
> James Cussens
> Room MVB 3.26
> Dept of Computer Science, University of Bristol
> Phone: +44 (0)117 455 8723
> https://jcussens.github.io/ <https://jcussens.github.io/>
> Funded PhDs available in Bristol in the following areas: Data Science
> <http://www.bristol.ac.uk/cdt/compass/>, Interactive AI
> <http://www.bristol.ac.uk/cdt/interactive-ai/>, Cyber Security
> <http://www.bristol.ac.uk/cdt/cyber-security/> or Digital Health
> <http://www.bristol.ac.uk/cdt/digital-health/>.
>
> _______________________________________________
> Scip mailing list
> Scip at zib.de
> https://listserv.zib.de/mailman/listinfo/scip
_______________________________________________
Scip mailing list
Scip at zib.de
https://listserv.zib.de/mailman/listinfo/scip
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.zib.de/pipermail/scip/attachments/20221028/b49eae2b/attachment.html>