Tutorials

These tutorials show many different ways to use FOMTools, following the same path a project usually takes: verify a first-order method, generate a fast implementation of it, and then drive that implementation from a CVXPY model. The notebooks are marimo files; their sources are in docs/tutorials on GitHub.

Sections

• Verification contains five notebooks that verify the worst-case performance of first-order methods on example problems: lasso, optimal control, network flow, nonnegative least squares, and portfolio optimization.
• Code generation works through the same five problems end to end, turning each verified method into an embedded C solver and timing it against direct Python. The code generation guide is the reference for the pipeline itself.
• CVXPYgen integration models a parametric problem in CVXPY and solves it with a generated FOM, combining CVXPYgen's canonicalization code with a FOMTools solver in one embedded C project.