lp.rhs seems incorrect about cylp HOT 3 OPEN

That will be a nice project! Thank you for posting this issue. I had a look and the problem seems to be with scaling. lp.rhs return the current rhs that the solver is working with (scaled). I added lp.variableScale and lp.constraintScale to the dev branch. So the un-scaled rhs can be computed from there. In your example variableScale=[100, 100] and it seems correct, doesn't it?

Anyways, do you think that lp.rhs should return the un-scaled rhs by default? Or is it fine the way it stands?

By the way if you pull the dev branch, you'll need to change the imports to reflect the recent package name change (CyLP->cylp).

from cylp.

I'm not sure I see the full picture of how you have things set up enough to answer your question definitively. The attribute lp.tableau seems to be computed with respect to the unscaled original matrix. As a user, I would assume lp.rhs matches lp.tableau (either both are scaled or both are unscaled). As a user, I would also prefer to have everything unscaled, since almost anything I can imagine doing with the data would require it to be unscaled first. I do really like the fact that you can easily find out what scaling was applied by the solver, though.

from cylp.

I agree. It makes more sense that way. Let me find the appropriate way to do it. Thank you!

from cylp.