osqp

Why we included this project

OSQP solves convex quadratic programs of the form minimize 0.5 x'Px + q'x under linear bounds, using an ADMM first-order method that stays fast on the sparse problems real-world models tend to produce. It is written in C, with bindings for Python, C/C++, and Julia, and the core solver can be embedded directly into larger applications rather than only called through a wrapper. That makes it a practical fit for model predictive control loops, portfolio optimization, and SVM training, where you often need a solution in milliseconds and can accept slightly less accuracy in exchange for speed. A companion modeling layer lets you state the problem as matrices and leaves the sparse linear algebra to the solver.