Diogo Ribeiro

Data Science & Research Lead writing practical notes on statistics, machine learning, forecasting, and data engineering. The useful posts here should add analysis, examples, code, experiments, or field judgment rather than paraphrase search results.

pinn-rk provides Runge-Kutta physics-informed neural networks with time-discrete losses in PyTorch. It focuses on Gauss, Radau and Lobatto style formulations for scientific machine learning problems where temporal discretization is part of the modelling choice.

Install

pip install pinn-rk

Package Metadata

  • Current release: 0.6.0
  • Requires Python: >=3.10,<3.13
  • License: MIT

Where It Fits

Use this package when the learning problem depends on time-stepping choices, numerical integration structure or a physics-informed loss that should be explicit and inspectable.