This folder contains the sources of the paper
Nonlinear model reduction on metric spaces. Application to one-dimensional conservative PDEs in Wasserstein spaces
by V. Ehrlacher, D. Lombardi, O. Mula et F.-X. Vialard.
You can also find some videos of reconstructed dynamics with our algorithms.
Python >= 3.7
Required modules: scipy, numpy, matplotlib, multiprocessing, cvxopt, itertools, os, sys, time, jsonpickle, pickle, argparse
The main file is test-all.py. To reproduce the results of the paper, run the command
python3 test-all.py -p <p> --id <id> --offline
where:
- <p> is the type of PDE problem (keys are: Burgers, ViscousBurgers, KdV, CamassaHolm)
- <p>\<id> is the filename of the folder where results are stored
- offline is an optional parameter to compute the offline phase
For instance, to reproduce the results on inviscous Burger's equation that are on the paper, run
python3 test-all.py -p Burgers --id paper --offline
Results will be stored in the folder Burgers/paper. The whole computation takes about an hour. For the other problems, the computational time is longer.
Copyright (c) 2019, Olga Mula (Paris Dauphine University).