TRANSFORMER PHYSX

Paper Supplementary Results

Additional visualizations from the numerical exampled tested in the journal paper.
Nick Geneva
Jun 17, 2021
It takes 2 minutes to read this article.

Supplementary Results

As a part of the the original paper, animated visualizations for each of the numerical examples are provided. For additional details on each numerical example, please refer to the paper.

Lorenz System

Prediction of ordinary differential equations from any intial state.

2D Navier-Stokes System

Transformer with Koopman based embeddings (Transformer-KM) is able to out perform embedding methods and ConvLSTM for prediction of flow around a cylinder at different Reynolds numbers.

Reynolds number 133
Reynolds number 433

3D Reaction Diffusion System

Prediction of 3D reaction diffusion system with random initial condition.