Hypatia.jl: a new solver for generic conic optimization

Hi friends, particularly those of you interested in conic optimization. We (myself and Lea Kapelevich, under the supervision of Juan Pablo Vielma) have just released v0.1.0 of Hypatia.jl, our pure-Julia open-source generic conic interior point solver. It is very much an alpha release for the most adventurous of optimizers, and the various low-level interfaces and options and defaults will change in future. It is accessible through a native interface and through MOI (hence JuMP), and passes the relevant MOI.Test tests. Currently the only documentation is code comments, though we have an extensive set of applied examples that are helpful for getting users oriented to modeling with “exotic” cones. Please see our early draft of a paper that explains the major motivation behind Hypatia and introduces some of the exotic cones and describes/benchmarks a subset of our examples. We would be very happy to have the JuMP community engaged in creating issues and PRs for Hypatia. After we have completed documentation and ironed out any egregious problems users might find and finished paper or two with benchmarks, we plan to make another announcement to encourage the less-adventurous users to give Hypatia a try.

24 Likes