bug-glpk
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Bug-glpk] numerical instability (cycling?)


From: Ali Baharev
Subject: [Bug-glpk] numerical instability (cycling?)
Date: Fri, 31 Jul 2009 23:40:14 +0200

Dear Andrew,

I suspect that i have ill-conditioned / badly scaled LP problems.

They should be either solved in a few iterations starting from a given
initial basis or the solver should return that the problem is
ill-conditioned / badly scaled and has given up the procedure.
Unfortunately my LP problems make the solver hang up, from the output
it seems to me that the simplex algorithm is cycling.

How can i dump these LP problems so that this phenomenon can be
reproduced on different machines?

I tried:

parm.msg_lev = GLP_MSG_ALL;
glp_term_out(GLP_ON);
glp_write_lp(lp, 0, "dump.lp");
glp_write_sol(lp, "dump.bas");

right before the critical call to glp_simplex.

These files and the output are here:

http://reliablecomputing.eu/glpk/dump.lp
http://reliablecomputing.eu/glpk/dump.bas
http://reliablecomputing.eu/glpk/console

As for the output, the objective consists of a single variable. The LP
problems are generated using the C API.

Many thanks,

Ali




reply via email to

[Prev in Thread] Current Thread [Next in Thread]