4 ms·

Good question, my thoughts: 1) to apply an LP or MILP to a practical problem that a business cares about requires a rare mix: one or a small number of people to

by jethkl 3y ago

Good question, my thoughts: 1) to apply an LP or MILP to a practical problem that a business cares about requires a rare mix: one or a small number of people to have domain knowledge, knowledge about LPs/MILPs, and a good fit to the problem at hand. 2) the types of problem where LPs/MILPs reduce business expense is (currently) different from the spaces where ML has found success. This could change, but LPs have been applied extensively to strategic/logistics/planning applications, which aren't as approachable as applications of chatGPT, xgboost, etc. 3) LPs - since they are convex and provide global solutions - don't naturally support a kaggle-type competition that pits individuals and teams against each other. 4) there exist good open source solvers and very nice APIs (scipy, cvxpy, cvxopt, etc for example) but also high-cost and high-performance commercial solvers that businesses do pay for (Gurobi, CPLEX, etc).

leethargo 3y ago
To add to the above: The high-performance commercial solvers typically offer a free (as in beer) licence for academics (students and researchers), so this subgroup has a smaller incentive to develop a competing solver. Similarly, researches who do spend their time implementing solver algorithms and running tedious computational experiements (the work that the software vendors put in) have historically had difficulties getting academic credit for their work, because the journals favored theoretical work. That being said, with HiGHS and SCIP, we have two open-source solvers developed in an academic setting, with a lot their graduates joining commercial software vendors. So it's not like these are two completely separate worlds.
- npalli 3y ago
  Sadly though, the Open source solvers seem to perform pretty badly compared to commercial solvers. https://plato.asu.edu/bench.html https://plato.asu.edu/bench.html
  - leethargo 3y ago
    True, especially on difficult problems. Open-source can be good enough for many practical applications, though. In my opinion, the gap in performance is less important, but the commercial offerings are typically more robust/reliable.

by jethkl 3y ago

leethargo 3y ago

To add to the above: The high-performance commercial solvers typically offer a free (as in beer) licence for academics (students and researchers), so this subgroup has a smaller incentive to develop a competing solver. Similarly, researches who do spend their time implementing solver algorithms and running tedious computational experiements (the work that the software vendors put in) have historically had difficulties getting academic credit for their work, because the journals favored theoretical work. That being said, with HiGHS and SCIP, we have two open-source solvers developed in an academic setting, with a lot their graduates joining commercial software vendors. So it's not like these are two completely separate worlds.

npalli 3y ago
Sadly though, the Open source solvers seem to perform pretty badly compared to commercial solvers. https://plato.asu.edu/bench.html https://plato.asu.edu/bench.html
- leethargo 3y ago
  True, especially on difficult problems. Open-source can be good enough for many practical applications, though. In my opinion, the gap in performance is less important, but the commercial offerings are typically more robust/reliable.