3 ms·
There are also ways around this by suitable generalizations of MILPs, namely mixed integer comic programs (MICPs), which are about as fast to solve in practice.
by LolWolf 7y ago
There are also ways around this by suitable generalizations of MILPs, namely mixed integer comic programs (MICPs), which are about as fast to solve in practice. This generalization allows you to use nonlinear (convex) functions in the objective and in constraints. Coupled with the integrality constraints, almost all nonlinear problems I’ve encountered can be written as MICPs.
- agravier 7y agomixed integer convex programming, not comic (unfortunately; although I hope someone takes this as a challenge). Or maybe you meant conic programming? As in MISOCP
- LolWolf 7y agoI did mean to write mixed integer convex optimization; on the other hand, conic programming includes convex programming as a special case (and vice versa in the case where cones are convex). Most convex problems we solve (e.g. When written in a high level DSL like CVXPY) are reduced to conic forms (such as SOCPs, but also SDPs, etc) before they are solved.
- LolWolf 7y agoOh, oops, I just realized I actually wrote "comic," oops! (My phone really does not like the word conic for whatever reason...) Would be a fun (and uhh, interesting?) challenge indeed, whatever that would entail :)
- kragen 7y agoIt fits in well with your username! There have been a number of (computer) programming environments that used a comic-strip-like format to depict a sequence of events; I think some are profiled in Watch What I Do and/or Your Wish Is My Command, which I don't have with me at the moment.