Hello everyone!
We are happy to announce that Convex.jl version 0.5 has been released! Since v0.4, we’ve seen some great additions and changes to the package.
Major changes:
Convex.jl can now solve optimization problems with complex variables and coefficients. We have added examples from AC power optimization, signal processing and quantum information theory to show the usefulness of our implementation. Some of them are:
- An application to power grid optimization
- Phase recovery from signals using relaxation similar to MaxCut
- Fidelity in Quantum Information Theory
Full documentation on how to solve complex-domain optimization problems with Convex.jl is available here. We’re excited to be the first DCP package in an open source language (and the second DCP package, after CVX in Matlab) to support complex-domain optimization.
Minor changes:
Enable precompilation
Added support for sparse matrices to quadform.
Distinguish between elementwise multiplication and matrix multiplication.
It’d be great for everyone to take it for a spin! We’d welcome pull requests with usage examples, particularly for complex-domain problems. Please file issues on the issue tracker with discovered bugs, ideas for how things should work, enhancements, and any issues with performance. Do try out the new functionality and feel free to ask questions.
Much of this has happened with the help of new and old contributors: we’re now up to 27 contributors and counting! Special thanks to @madeleineudell, @mlubin,@dvij, @tkelman, @petterwittek, @thomasschiet, and @ayush-iitkgp.
Cheers!