Contributing

There are many ways you can contribute to SCS, and we welcome all help! Here are some ideas (of varying difficulty):

• Answer an open GitHub issue

• Improve the documentation (the website or in the code)

• Add more Examples to the docs

• Improve the test coverage

• Improve the Anderson acceleration stability

• Improve the data equilibration

• Improve the heuristic re-scaling

• Determine how to select the TAU_FACTOR term

• Add other new linear system solvers

• Refactor the linear solvers to only compile a single binary with all solvers

• Add interfaces for other languages (or improve the current interfaces)

If you are interested in helping out, please start by opening a GitHub issue so we can track progress and ensure that our priorities align.

Building the docs

The example pages show real solver output, which is generated at build time rather than committed, so it can never drift from the solver. Building the docs therefore runs the C and Python examples and needs their dependencies:

pip install sphinx sphinx-rtd-theme breathe docutils   # the docs themselves
pip install scs numpy scipy cvxpy                      # to run the examples
cd docs/src && make docs

A C compiler and BLAS/LAPACK are also required, for the C example. Use make example_outputs to regenerate just the captured output, and DOCS_PYTHON=/path/to/venv/bin/python to point at a specific interpreter.

The one exception is examples/qp.m.out, which is committed: refreshing it needs a MATLAB licence, so it is updated by hand.