[Bug-glpk] Re: feasible LP declared infeasible by GLPK

[Andrew Makhorin]

I looked at your problem more attentively and found that first 58
columns have very large negative lower bounds (-1e8). If these columns
can reach their lower bounds, your problem seems to be badly scaled.
If not, it is desirable to specify more appropriate bounds which are
not such large (in the optimal point of the lp relaxation they all are
basic and have zero activity). If you mean these columns have no lower
bounds, lower bounds must not be specified at all.

Andrew Makhorin