simulation.utils.machine_learning.models.test.test_wgan_critic module

Perform some basic tests for the WGAN critic.

Summary

Functions:

main
test_forward
test_optimization_step	Testing very basic functionality of optimizing.
test_weight_clipping

Reference

test_weight_clipping() → None

test_forward() → None

test_optimization_step()

Testing very basic functionality of optimizing.

Testing if the optimization works is hard. Here, some very basic things are tested:

• Does the wasserstein distance increase when running the optimization?

• Is the distance close to zero, if random distributions are given

  and the generator is the identity?

• Is the generator unchanged?

main()