Name		Name	Last commit message	Last commit date
parent directory ..
README.md		README.md
baseplots.py		baseplots.py
bayes.py		bayes.py
config.py		config.py
config_heat.py		config_heat.py
experiments.sh		experiments.sh
gpkernels.py		gpkernels.py
main.py		main.py
pde_models.py		pde_models.py
plots_paper.py		plots_paper.py
step1_generate_data.py		step1_generate_data.py
step2_fitgps.py		step2_fitgps.py
step3_estimate.py		step3_estimate.py
step4_plot.py		step4_plot.py
utils.py		utils.py
wlstsq.py		wlstsq.py

README.md

GP-BayesOpInf for PDEs with Multiple Training Trajectory

This folder implements GP-BayesOpInf for problems in which multiple trajectories of training data are available.

Main Routines

main.py: run an experiment and show/save results.
step1_generate_data.py: generate experimental snapshot data.
step3_estimate.py: solve the Bayesian operator inference problem, selecting the regularization hyperparameters automatically. Construct the posterior.
step4_plot.py: plotting tools.

Links

step2_fitgps.py: fit Gaussian processes to data (same as the single-trajectory case).
baseplots.py: utilities for plotting experiment results.
bayes.py: Bayesian posterior models.
gpkernels.py: GP kernels and regression matrices.
pde_models.py: full-order partial differential equations models.
wlstsq.py: weighted least-squares solvers.

Utilities

config.py: configure settings (matplotlib, file/folder names, etc.).
config_heat.py: simulation scenario for the heat equation problem.
utils.py: define tools for loading data and saving figures.

Reports

runlog.sh: Log of commands used to run the experiments reported in the paper.

Data

These folders are not tracked by git, but may be generated by running the code.

data/: experiment results for visualizations.
figures/: saved figures, organized by date.