Sliced Distribution Matching based on Cumulative Distribution Functions with Applications to Control
This repo accompanies the paper by the same name. We introduce a sliced distance between distributions based on the discrepancy of the CDFs of one-dimensional linear projections.
The code contained in the repo was run to obtain the results in the paper and is structured as follows.
The results of the two-sample test section can be obtained by running two_sample_test.py.
The folder /distribution_steering contains the experiments for the distribution steering application. To replicate our results, please run python distribution_steering/main.py.
The folder /ergodiccontains the experiments for the ergodic control application. To replicate our results, please run python ergodic/main.py.
Figure 1 of the paper was obtained by running overview_fig.py.