Basis launches Dynestyx, a probabilistic programming library for dynamical systems

xuanalogue · x · 2026-11-11

Basis has released Dynestyx, a probabilistic programming library for dynamical systems, with an accompanying paper and code.

It targets a real pain point: researchers in fields from epidemiology to turbulent flows often have to rewrite substantial code when changing models or inference methods. Dynestyx offers a shared interface and sensible defaults so analyses can evolve without being rebuilt.
It supports four core tasks: forecasting, state inference from noisy partial measurements, system identification, and counterfactual/intervention reasoning.
The work is part of Basis's broader platform of modeling and reasoning technologies, aiming to include models verified via formal methods.

More from Research