[Clp] Objective is nan (sometimes assertion failure)

[Sébastien Roy]

This usually means your linear program is unbounded so the objective
function value is infinity. If your modeling a problem that should bound
you must have an error in your linear program model.

On Sat, May 6, 2017 at 6:20 PM, Aleksandr M. Kazachkov <akazachk at cmu.edu>
wrote:

> Hi everyone,
>
> I am not sure if there is something wrong with my setup, or if this is a
> bug, but I would appreciate any help in either case. Thank you in advance!
>
> I have Clp version 1.16 set up on a Fedora machine. The compilation
> options are: enable_debug=yes, enable_gnu_packages=yes, and I am using blas
> and lapack. There are no problems when I run "make test".
>
> However, on the following LP (which solves fine when using Bcp-1.4.3,
> which maybe is using an older version of Clp), when presolve (initial) is
> off, I get that the objective value is "nan". In the context this LP came
> up (but not when I saved it and ran it on its own), I got the following
> assertion failure:
> ../../../Clp/src/ClpSimplexDual.cpp:3216: int
> ClpSimplexDual::changeBounds(int, CoinIndexedVector*, double&): Assertion
> `!("should not be here")' failed.
>
> Specifically, I get the error after the following two lines:
>   solver.setHintParam(OsiDoPresolveInInitial, false);
>   solver.initialSolve();
>
> _______________________________________________
> Clp mailing list
> Clp at list.coin-or.org
> https://urldefense.proofpoint.com/v2/url?u=https-3A__list.
> coin-2Dor.org_mailman_listinfo_clp&d=DwICAg&c=Ngd-
> ta5yRYsqeUsEDgxhcqsYYY1Xs5ogLxWPA_2Wlc4&r=QucHtiOE6l_
> F6qqZBx1wRnOlfxGN76vTLMBmdvbxugw&m=3px0r_GTXRracL5O_
> VCNIs6wypj5AB2PkAsFdkd9NpA&s=cIujfK_WHj04A3bZYaw7BY3otKyFlEVCPn7yL
> d1siEc&e=
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/clp/attachments/20170506/a452afc0/attachment.html>