Thanks for using the package and being interested in contributing to the project. There are several ways to contribute.
We are mostly following the PEP 8 python style guide with line width 80 and Google style docstrings. At parts we diverge from these styles and could probably need some help to find back to it. Contributions to a more extensive and better documentation are always welcome. In similar vein, the tutorials can be improved with more usage capabilities and more interesting examples.
If you find any bugs or unexpected behaviour, please open an issue. Include a minimal working (or not working) example following above mentioned style guidelines. If you already have a workaround for a bug/issue, open a pull request. Fork the repository and run the testing scripts in order to check that the code doesn't break unexpectedly.
Our current testing scripts are rather bare bone. A more extensive testing suit will improve the package.
The different modules often contain various little TODO-tags, which indicate possible performance improvements, clean-up, or additional features. We will be grateful, when someone ticks them off the list.
There are several more features planned for the future. Those include in no particular order:
- Adding more general SU(3)-structure learning as in 2021.04656.
- Adding (vanishing) divisor integrals to fix the metric to particular points in Kaehler moduli space for other Ansätze than the PhiModels.
- Adding MCMC-methods as comparison to the theorem by Shiffman and Zelditch.
- Work out generalisation of [SZ] for the Toric manifolds.
- Improve performance of point generation. Either by adding other backend options to faster libraries or utilizing GPU, or ... ?
- Adding toric interface to utilize triangulations and toric data from cytools.
- Add more potential methods, such as energy functionals, learning the hbalanced metric, learning function basis.
- Add bundle metric.
We are of course open for more suggestions and welcome every feedback.