All Tools
O
Dev ToolsFreeOpen Source
OPTIMISTIX
Nonlinear root-finding and least squares in JAX
Apache-2.0
ABOUT
Nested scientific models need root-finding and nonlinear least squares that stay in JAX, but SciPy optimize drops off the accelerator. Optimistix provides those solvers so Equinox loops differentiate through an inner fit.
INSTALL
pip install optimistixINTEGRATION GUIDE
1. Nest a nonlinear least-squares fit inside a JAX training loop
2. Solve a root-finding problem with a differentiable JAX solver
3. Minimise a scientific residual without leaving Equinox
4. Replace a SciPy optimize call that copied arrays off GPU
TAGS
pythonjaxoptimizationequinoxleast-squaresroot-finding