This Python package provides various ingredients for research in polynomial optimization.
The leading design principle is providing just the right set of tools to quickly and intuitively sketch algorithms and perform experiments.
For instance, while the package doesn't provide a fully-featured SDP solver, it does provide all routines needed to write one in a few lines (e.g. parsing SDP data from a file, PD line search, dealing with block matrices, etc.).
See examples/ and the section below for various showcases.
To get an overview over current set of features check out the docs.
I develop the core library as needed by the particular algorithms or methods from the literature that I'm interested in.
Each item in the following list corresponds to one such showcase in the examples/ directory.
- [WIP]
bad-csdp-clone: This is a simple re-implementation of gh:coin-or/Csdp with focus on code readability. The solver comes with a nice CLI and is also used in other examples.
Usually I focus on developing examples.
If they turn out to be useful and stable enough, I'll integrate them into the main package.
Because of this, semantic versioning of the package doesn't make a lot of sense, so I'm using a year.month.day version format instead.
Whenever a considerable amount of progress has been done and the library has reached a stable state I'll bump the version.