Time-series learning of latent-space dynamics for reduced-order model closure
R. Maulik, A. Mohan, B. Lusch, S. Madireddy, P. Balaprakash, D. Livescu
https://doi.org/10.1016/j.physd.2020.132368
These are results using the hand-tuned hyperparameters
These are results using deephyper for hyperparameter optimization
For visualizing results from LSTM deployments use python Burgers_POD_ROM.py
and for NODE deployments use python NODE_Burgers.py
. Change deployment_mode = 'test'
to deployment_mode = 'train'
within these files to retrain the framework. This repo will have further documentation added to it slowly - in the meanwhile please contact rmaulik@anl.gov for any questions.