osqp VS csips

There's also two bindings for the osqp library (which is written in C), osqp published 2 years ago and osqp-rust published 3 months ago. I don't know what are the differences between them, but they both target osqp 0.6.2 (released in 2021) while the last released version is osqp 0.6.3 which was released last week.

Cvxpy is overkill for a standard quadratic program. I’d recommend trying OSQP https://osqp.org which can take advantage of sparsity.

I have been using OSQP [1] quite a bit in a project where I needed to solve many quadratic programs (QPs). When I started the project, OSQP didn't exist yet; I ended up using both cvxopt and MOSEK; both were frustratingly slow.

After I picked up the project again a year later, I stumbled across the then new OSQP. OSQP blew both cvxopt and MOSEK out of the water (up to 10 times faster) in terms of speed and quality of the solutions. Plus the C interface was quite easy to use and super easy (as far as numerics C code goes) to integrate into my larger project.

[1] https://osqp.org/

For quadratic programming—which is a class of problems in convex optimization, which is a sub-field of numerical optimization in general—a solver that is frequently used is OSQP. Although it is implemented in C++ you can also use it in Python thanks to its bindings. If your goal is to use a solver that's state-of-the-art and relatively versatile it is a good pick. If your goal is to find the best solver for a given problem, then there is no one-stop-shop. For example in this benchmark OSQP was the best-performing solver for sparse problems but quadprog performed better on dense problems.

I actually just finished implementing an extremely simple Integer Linear Program solver in Python as an educational exercise, wrapping scipy's linprog function to solve the linear relaxation. It has an expression syntax so you don't have to specify the matrix and vectors for the standard form, and it does branch-and-cut on the linear relaxation

https://github.com/cwpearson/csips