Sven Verdoolaege a écrit :
On Fri, Oct 30, 2009 at 01:31:14PM +0100, Francois Galea wrote:
Thanks, I will have a look at these commits once I have implemented the support for bignums in the PPL solver.
Do you mean "big parameter"?
Indeed, this is what I meant.
There's nothing really interesting to see in those commits. I just mentioned them to show I'm not hallucinating. In any case, in piplib, I just implemented the tricks described in the manual. You should probably look at that before you start implementing. Note that the manual proposes two different tricks to handle negative unknowns and negative parameters. I have no idea why. I mean, it's clear why you shouldn't use the other trick in the main tableau, but it's not clear why you shouldn't use a big parameter in the context tableau. But you are problaby not using the dual simplex + Gomory cuts technique in the context, so you shouldn't have a problem with negative parameters.
I will have a further look at all of this soon. Meanwhile, I still have fixes to do on the solver core.
Thank you for all your advice.
Best,
François.