Implicit Rank Minimizing Autoencoder with Weight-Decay:
This repo contains a demo for Implicit Rank Minimizing Autoencoder with Weight-Decay (IRMAE-WD) implemented in pytorch applied to the KSE, L=22 dataset. This repo accompanies the work outlined in the following paper: K. Zeng et al 2024 Mach. Learn.: Sci. Technol. 5 025053
DManD for Couette flow:
This repo contains the data-driven manifold dynamics (DManD) model described in A.J. Linot and M.D. Graham, J. Fluid Mech. 2023 . Here we provide POD modes for minimal flow unit turbulent Couette flow at Re=400, a hybrid autoencoder that reduces the dimension down to 18 degrees of freedom, and a neural ODE that evolves this 18 dimensional system.
KSE reduced order models and neural ODEs:
This repository contains code for training autoencoders on data generated from the Kuramoto-Sivashinsky Equation (KSE) at a domain size of L=22, and code for running short trajectories of the reduced order models at domain sizes of L=22, 44, and 66 described in A.J. Linot and M.D. Graham, Chaos 32.7 (2022).
Data-driven discovery of intrinsic dynamics:
This distribution contains code that implements an atlas of charts in the context of data-driven forecasting of dynamical systems, as described in “Data-driven discovery of intrinsic dynamics,” by D. Floryan and M. D. Graham, Nat Mach Intell 4, 1113–1120 (2022).
Data-Driven Wavelet Decomposition (DDWD):
This distribution contains the code needed to create data-driven wavelets, as described in “Discovering multiscale and self-similar structure with data-driven wavelets,” by D. Floryan and M. D. Graham, PNAS, 2021.