SuperSCS: Fast & Accurate conic programming

About SuperSCS

SuperSCS is a fast solver for conic problems of the form

\begin{eqnarray*} &&\mathrm{Minimize}\ c' x \\ &&Ax + s = b\\ &&s\in\mathcal{K}, \end{eqnarray*}

where \(A\in\mathbb{R}^{m\times n}\) is a (sparse) matrix, and \(\mathcal{K}\) is a closed, convex, proper cone.

\(\mathcal{K}\) can be the Cartesian product of the zero cone, the positive orthant, the second-order cone, the positive semidefinite cone and many other.

SuperSCS is based on the algorithmic scheme SuperMann applied to a Douglas-Rachford splitting on the self-dual homogeneous embedding of the original problem.

SuperSCS achieves higher accuracy and faster convergence compared to SCS as you can see for example in the CVX examples page.

Documentation

1. Installation
   • Detailed installation instructions
2. Documentation
   • Cones
   • Using SuperSCS
     • in C
     • in MATLAB
     • in Python
     • via CVX (MATLAB)
     • via CVX (Python)
   • Examples
     • Examples in C
     • Examples via CVX
   • Saving/Loading Problems (YAML)
   • Logging progress data
3. Performance
   • Benchmarks
   • Profiling
4. Under the hood
   • About the algorithm
   • Residuals and termination
   • Quasi-Newtonian directions
   • Default Settings

Download

The source code of SuperSCS is available on github.

Check out the installation guide.

Verification

Quality assurance:

• unit tests (make run-test)
• memory management tests using valgrind
• continuous integration (Travis CI) (Linux + Mac OSX)
• continuous integration (Appveyor) (Windows)
• coverage reports on Codecov
• code quality reports on codacy
• lcov report using make cov COV=1

Contributing

SuperSCS is an open-source project to which you may contribute.

Before contributing or filing an issue, please read this guide.

License

SuperSCS is licensed with an MIT license.

The MIT license is a short and simple permissive license with conditions only requiring preservation of copyright and license notices.

Licensed works, modifications, and larger works may be distributed under different terms and without source code.

You only need to keep LICENSE.txt and, if you redistribute the source code, the copyright notices therein.

Do you have a question?

Feel free to reach us on Gitter.

Click here to chat with us:

Cite SuperSCS

SuperSCS is based on the algorithmic scheme presented in:

A. Themelis and P. Patrinos, SuperMann: a superlinearly convergent algorithm for finding fixed points of nonexpansive operators, arXiv:1609.06955, 2017.