Software

jaxmat

🚀 jaxmat is an open-source Python library for implementing material constitutive models in JAX.

jaxmat accelerates the development of new material models by combining concise syntax, reusable building blocks, and automatic differentiation, eliminating the need to manually derive consistent tangent operators. It also provides a natural foundation for data-driven constitutive modeling.

• Bridges the gap between classical constitutive modelling and modern ML frameworks
• Automatic differentiation replaces hand-derived jacobians and tangent operators
• jax.jit and jax.vmap give GPU-accelerated, vectorised evaluation across thousands of material points with a single line of code
• Modular design: swap yield surfaces, hardening laws, or flow rules independently, or replace any component with a neural network
• Every parameter is differentiable, making material identification and inverse problems natural to formulate

dolfinx_materials

dolfinx_materials is a Python add-on package to the dolfinx interface to the FEniCSx project. It enables the user to define complex material constitutive behaviours which are not expressible using classical UFL operators.