> > I have a fairly trivial LP problem that I'm trying to > solve using GLPK, > but it says there's no feasible-point even though I can compute one by > hand. Here is the coefficient matrix: > > 1 0 -0.5055 0.0000 0.0000 0 0 0 0 0 0 > 0 1 -0.4207 0.0000 0.0000 0 0 0 0 0 0 > 1 0 0.0000 -0.4779 0.0000 0 0 0 0 0 0 > 0 1 0.0000 -0.3355 0.0000 0 0 0 0 0 0 > 1 0 0.0000 0.0000 -0.4023 0 0 0 0 0 0 > 0 1 0.0000 0.0000 -0.1920 0 0 0 0 0 0 > 1 0 -252.9900 0.0000 0.0000 1 0 0 0 0 0 > 0 1 -210.5600 0.0000 0.0000 0 1 0 0 0 0 > 1 0 0.0000 -239.1900 0.0000 0 0 1 0 0 0 > 0 1 0.0000 -167.9000 0.0000 0 0 0 1 0 0 > 1 0 0.0000 0.0000 -201.3400 0 0 0 0 1 0 > 0 1 0.0000 0.0000 -96.0900 0 0 0 0 0 1 > > The first six rows are set as >=0.0 and the second six are set as > =0.0. > I'm minimizing the sum of the last six column-variables, that absorb > the "residue" of what might otherwise make the last six rows nonzero. > All variables are set as >=0.0. An example of a feasible-point is the > following, in order of the columns they mutliply with: > > 201.34 > 96.09 > 0.796 > 0.842 > 1 > 0.05 > 89.84 > 0.07 > 53.78 > 0 > 0 > > Is this matrix degenerate in some way that breaks GLPK? I've > tried the > various combinations of lpx_simplex, lpx_interior, lpx_scale_prob, and > LPX_K_PRESOL=1. I assume that GLPK is working as designed, since the > "make check" passes okay. I'm running on an IBM Thinkpad > running RHEL 4
Please write the problem data to a text file, say, in mps or cplex format (lpx_write_mps, lpx_write_cpxlp) and post it to me. Andrew Makhorin _______________________________________________ Help-glpk mailing list Help-glpk@gnu.org http://lists.gnu.org/mailman/listinfo/help-glpk