Ask HN: Do you use an optimization solver? Which one? Do you like it?

What I saw often, and this was in finance, that one or two developers gain proficiency in converting problems to a specific solver tool, and then we could never move to a different library because it would have been so much effort to relearn.

1. At least they exist! They give a good rough idea of the relative performance of the various solvers. Also, I think it provides a sense of competition that is great for innovation. Some of the newcomers push press releases about having "won a prize" when they leapfrog the competition in one of the (sometimes weekly) benchmarks.

2. For the MIP benchmarks [2], a lot of thought has gone into doing them properly [3].

Don't be scared by the "End of a benchmarking era" banner. In 2018, Gurobi cherry-picked numbers that made them look good in some promotional material, while suggesting endorsement by Hans Mittelmann. There was a whole fuss about it and they had to publicly apologize [4]. This was a very dumb move since they were the fastest solver anyways, just not by as much as they claimed. CPLEX and Xpress took this as an excuse to take their toys and go home (they have since refused to participate in benchmarking). Gurobi was banned from the benchmarks for a while, but now they're back.

[1] http://plato.asu.edu/bench.html

[2] http://plato.asu.edu/ftp/milp.html

[3] https://link.springer.com/article/10.1007/s12532-020-00194-3

[4] https://web.archive.org/web/20181224083542/http://www.gurobi...

At one time, OR looked good to me, and I bet a lot of my career on it: Off and on I heard about various parts of optimization. Then at FedEx, in a rush I wrote some simple software that produced nice fleet schedules and literally saved the company. Smith's remark "solved the most important problem ...".

But since airplanes are staggeringly expensive, I considered optimization. That looked like 0-1 integer linear programming with mostly just one row for each city to be served and one column for each candidate single airplane tour from Memphis and back. In generating the candidate tours, could easily handle lots of complicated costing and really goofy constraints. The optimization step itself was just set covering.

But my wife was still in Maryland; the promised stock was 18+ months late; I went and got an applied math Ph.D. with a lot of emphasis on optimization -- dissertation in stochastic optimal control. I taught linear programming in a business school (trying to help my wife recover from the stress of her Ph.D. -- she never recovered and her body was found floating in a lake; the two Ph.D. degrees were very costly).

Then, after sending 1000+ resumes, net, I had to conclude: Nearly no one in business gives as much as even a small fart about optimization or operations research.

I did stumble onto some small projects: One of these was a marketing resource allocation problem, 0-1 integer linear programming, 600,000 variables, 40,000 constraints. I used the IBM OSL (Optimization Subroutine Library), called from some Fortran code I wrote, and used some Lagrangian relaxation -- in 900 seconds of computing, with as I recall 500 primal-dual iterations, got a feasible solution, from the Lagrange multiplier bounding, within 0.025% of optimality. The customer was not much interested: The CEO had a buddy, his CTO, and my success had apparently embarrassed the CTO and, thus, displeased the CEO. I never talked with the CTO after my success, but his belief was that his simulated annealing, running for days, would be the best approach. So, he was torqued at my success.

Had a similar situation in another problem.

I had to conclude: Optimization seems like an obviously valuable step, tool, and often quite worthwhile. E.g., for some $100 million project, optmization might save 10%, $10 million.

There were some Nobel prizes based on optimization.

There have been lots of successful, valuable, optimization projects reported.

So, there should be plenty of people in business eager to pursue optimization? Right? I assumed so. I was wrong.

But the guys who have the money and the responsibility and make the decisions very much don't trust or like operations research or optimization. They see such a project as a lose-lose situation: If the project fails, and some will, or just does not do very well, then their career takes a hit. If the project is nicely successful, then that success is a force in the organization that the C-level (CEO, CTO, CFO, etc.) can't really control. The C-level doesn't like to be out of control. Instead, they want to do things their way, and that does not include operations research.

So, I gave up on operations research and optimization.

So, instead of trying to have a career, doing optimization, working for a salary for people who don't want optimization, I'm doing a start-up. There is some pure/applied old/original math in the core of it, but no users will suspect anything mathematical.

This would be something in the objective function, I think? Assign some daily cost to having stock in inventory and a penalty per day that the "supply delivery to customer" task completes after the deadline.

IIRC ILOG solver used to have a lot of constraints that made implementing that kind of thing easier.