Funding push for new open-source MILP solver HiGHS

We are pleased to announce a funding push for the new kid on the open-source MILP solver block, HiGHS:

https://pypsa-meets-africa.github.io/highs

Please donate generously and tap your donor networks for bigger sums! We’re aiming to raise $114,000 in the first year for the developer team at the University of Edinburgh led by @JulianHall, with more to follow. There is a funky PDF for sharing with potential donors.

I’m not sure this crowd needs the hard-sell on this, but here we go:

As a community we have made lots of progress in developing open energy models (of which there may now be too many…) and some progress in opening up important energy datasets. But the missing piece has always been solvers. Academics can get licences for commercial solvers, but NGOs, non-NGO interest groups, companies, ministries and even academics running interactive websites cannot afford licences running up to USD 45,000 a year. Unfortunately established open solvers like GLPK and cbc don’t cut it for large energy system problems. Beyond cost, there is obviously a big benefit for the optimisation community to have insight into how the code works and develop it further.

HiGHS, led by a research team around @JulianHall at Uni Edinburgh, has entered the scene and stolen the show with performance that gives Gurobi a run for its money:

Based on discussions with the developer team, they reckon there is some lowish-hanging fruit to boost the performance of HiGHS for energy system problems. In response, @MaxParzen has led this effort to coordinate a funding push, with some help from @JulianHall, @JesseJenkins and me, as well as from other members of the modelling community who have offered “impact statements”.

Thanks for helping us boost this effort - the whole energy community will benefit from performant open source solvers, and probably many other communities doing optimisation too!

2 Likes