PhySO

Why we included this project

When you have measured data and suspect there's a clean analytical law underneath, PhySO is worth a look. It searches functional forms with deep reinforcement learning and returns a compact symbolic expression, so instead of a black box you get an actual equation you can inspect and reason about. Its physics focus shows in dimensional analysis, which enforces unit consistency and cuts the search space, and in a class-based mode that fits a single functional form to several datasets at once, each with its own parameters. That's handy for physicists and engineers working with oscillators, dynamical systems, or empirical scaling laws. If you want interpretable results and don't mind tuning hyperparameters, it's a solid fit for exploratory analysis.