opifex 0.1.0

Maiden release of the Opifex Computational Framework — a JAX-native platform for scientific machine learning.

What's in this release

• Neural operators (FNO, TFNO, DeepONet, GNO, PINO, UQNO, LocalFNO, SFNO, UNO)
• Physics-informed neural networks across Poisson, Burgers, Navier-Stokes, Wave, Helmholtz, Allen-Cahn, diffusion-reaction, advection, Euler beam, and inverse-diffusion problems
• Bayesian and uncertainty-quantification operators with epistemic/aleatoric decomposition
• Neural DFT and learned XC functionals
• Meta-learning (MAML, Reptile) and L2O on parametric PDEs
• Advanced training: NTK analysis, GradNorm loss balancing, RAR-D adaptive sampling
• Benchmarking framework with PDEBench integration, statistical analysis, and CI guard
• Distributed training utilities (DeviceMeshManager, SPMD, sharding rules)
• Scientific data loaders (Darcy, Burgers, diffusion, shallow water, Navier-Stokes, PDEBench HDF5, VTK meshes)

Install

pip install opifex

Documentation

See RELEASING.md, CHANGELOG.md, and the docs site for the API reference and example walkthroughs.