OpenSourceEconomics / lcm

Solution and simulation of life cycle models in Python with GPU acceleration.
Apache License 2.0
16 stars 1 forks source link

Fully discrete Model #26

Closed mo2561057 closed 1 year ago

mo2561057 commented 1 year ago

Solving and Simulating a fully discrete model works without error and produces reasonable results. Function

Function that calculates conditional continuation policy ( https://github.com/OpenSourceEconomics/lcm/blob/8c46c3f414b38a7fcf514654e184f8bf9b3a3b4b/src/lcm/entry_point.py#L222-L262) however behaves odd.

If all the discrete values are scalars it trows an error and if some values are array it takes the argmax over a discrete axis. (See this test: https://github.com/OpenSourceEconomics/lcm/blob/b0574b44344f18eb7f0a395cf5ece40f17f18b85/tests/test_entry_point.py#L266-L306) I have included a test that fails for now.

timmens commented 1 year ago

Thanks a lot for opening the PR and for writing the main part of the tests. I merged PR #47 into main beforehand, which enables discrete models. Merging this PR into main for the additional tests.