pina

What problem does it solve?

Provides a practical framework to build physics-informed neural networks (PINNs) and neural operators for solving PDEs, inverse problems, and operator learning with PyTorch.

Core Features & Use Cases

• PINN and neural-operator solvers for forward and inverse PDE problems.
• Problem definition, model construction, training with MLflow integration, and visualization.
• Use cases include barrier problems in engineering, parameter identification, and multi-query operator learning.

Quick Start

Train a PINA model by defining a PDE problem, choosing a network, wrapping it in a PINN solver, and running a training loop to obtain predictions and diagnostics.