RNN architectures trained with Backpropagation and Reservoir Computing (RC) methods for forecasting high-dimensional chaotic dynamical systems.

Runge-Kutta adaptive-step solvers for nonlinear PDEs. Solvers include both exponential time differencing and integrating factor methods.

Framework to learn effective dynamics and couple a macro scale simulator with a fast neural network latent propagator.

An AOT-based algorithm to estimate multiple unknown parameters in the Kuramoto-Saviashinski equation. Source code for the paper "Concurrent Multiparameter Learning Demonstrated on the Kuramoto-Sivashinsky Equation" by Pachev, Whitehead, and McQuarrie.

Numerical Evidence for Sample Efficiency of Model-Based over Model-Free Reinforcement Learning Control of Partial Differential Equations [ECC'24]

Spectral Integration and Differentiation Algorithms. Includes FFTs, Chebyshev Transforms, and Hankel transforms. Exponential time differencing and integrating factor Runge-Kutta methods.

A Python package to simulate and measure chaotic dynamical systems.

pseudospectral (fourier) solutions of a few 1-dimensional PDEs

Code for the Modeling From Measurements report