The internal LP solver is a bit slow, at least for me using pygambit.nash.lp_solve. In an experiment solving zero-sum extensive form games with ~100 infosets for each player, the internally linked solver takes ~18s, compared to lp_solve taking ~0.1s for the same LP constructed manually.
So, I'd like a feature in pygambit to use a different LP solver. (I see that by passing external=True I can use Gambit's external command line solver. Apologies if there's already a way to use this argument to call an LP solver of my choice.)
The internal LP solver is a bit slow, at least for me using pygambit.nash.lp_solve. In an experiment solving zero-sum extensive form games with ~100 infosets for each player, the internally linked solver takes ~18s, compared to lp_solve taking ~0.1s for the same LP constructed manually.
So, I'd like a feature in pygambit to use a different LP solver. (I see that by passing external=True I can use Gambit's external command line solver. Apologies if there's already a way to use this argument to call an LP solver of my choice.)