Full Multi-Solver RL Environments Release

[ludgerpaehler]

This release extends HydroGym from a single-solver backend with the potential for multiple solvers to a native multi-solver reinforcement learning framework for flow control including environments in:

• Firedrake
• Jaxfluids
• m-AIA
• Nek5000

In addition this release introduces the data manager abstraction which handles the environment's data natively, and across solver backends. The accompanying environment data is stored in the corresponding HuggingFace dataset repository which is auto-accessed by the data manager.

The documentation has seen a complete rewrite from the previous readthedocs version and now resides at dynamicslab.github.io/hydrogym/.

---

This discussion was created from the release Full Multi-Solver RL Environments Release.