-------- Original Message -------- Subject: Re: [Bug-glpk] Bug in glpk 4.21? Date: Sat, 1 Sep 2007 01:54:28 +0400 From: Andrew Makhorin mao@gnu.org Reply-To: Andrew Makhorin mao@gnu.org To: Roberto Bagnara bagnara@cs.unipr.it CC: bug-glpk@gnu.org References: 10376349.20070831204809@gnu.org

For the attached testcase, we obtain that

$ glpsol p0033.mps --max -o output.gomory --gomory

gives an optimum of 5201, while

$ glpsol p0033.mps --max -o output.intopt --intopt

gives an optimum of 5131. The optimum 5201 is confirmed by a completely different MIP solver. Does this behavior indicate there is a bug somewhere?

The bug appears if the constant term of the objective is non-zero either originally or due to fixing some columns at non-zero value by the mip preprocessor.

In my previous post I just commented out the corresponding fragment leading to wrong results.

To fix the bug in a correct way please replace lines 748 and 751 (the routine ios_preprocess_node, file src/glpios02.c):

748: L[0] = -DBL_MAX, U[0] = mip->mip_obj;

751: L[0] = mip->mip_obj, U[0] = +DBL_MAX;

by the following ones:

748: L[0] = -DBL_MAX, U[0] = mip->mip_obj - mip->c0;

751: L[0] = mip->mip_obj - mip->c0, U[0] = +DBL_MAX;

Please note that these changes will appear in the next version of the package.

Andrew Makhorin