Scientific ML
GitHub ↗GPry
Gaussian-process surrogate modelling and active learning for expensive Bayesian inference.
Tools built around a recurring idea: make difficult computations tractable without hiding what the machinery is doing.
Gaussian-process surrogate modelling and active learning for expensive Bayesian inference.
Simulation and inference for scalar-induced gravitational waves and primordial curvature perturbations.
High-throughput, JAX-backed simulations of supermassive black-hole binary populations and their nanohertz signals.
A unified bridge from Discovery PTA models to Eryn, nessai, JAX-NS, and GPry.
Typed graph planning with deterministic compilation, validation, execution, and local repair.