Re: [Help-glpk] GLPSOL for Windows & docs in PDF

[Andrew Makhorin]

> my result for 'qui' problem from MIPLIB 3.0 is 16h10m22s on Intel 3.0
> GHz (and of course 'air05' - 4h51s, 'mod011' - 11h49m17s :).
> Solved to optimum with default options.

> Can anybody find/show something with longer/bigger runtime from the
> abovementioned testkit?

The benchmarks above are still hard for glpk :(

Note also that by default the glpk mip solver does not use cutting
planes that may be inappropriate in many cases. The situation is the
same as for p0548. Old versions of the glpk mip solver based on the
branch-and-bound could not find even an integer feasible solution
for an hour while glpk 4.34 using the pseudo-cost branching and mir
cuts solves it to optimality for 3 seconds.