search

danielapai / bioverse

A simulation framework to assess the statistical power of future biosignature surveys
MIT License
7 stars 5 forks source link

Hypothesis grids crash when empty samples occur #46

Open matiscke opened 6 months ago

matiscke commented 6 months ago

In the case where a grid of parameters in a test_hypothesis_grid leads to an empty sample, the run gets stuck*.

*progress bar doesn't progress anymore, task keeps running until interrupted.

astral-carrier commented 5 months ago

I had my progress bar hang for nearly 20 minutes, but then it unfroze and immediately jumped to 100%.

It took about 39 minutes in total to run code cell 11 on Example1.ipynb, which is in line with my previous runs.

I suspect the issue is the progress bar being inaccurate, as opposed to a bug with test_hypothesis_grid. How long did you wait on your task before interrupting it?

matiscke commented 5 months ago

I did try smaller grids that should finish within minutes. With certain choices of parameter range it got stuck, with others it finished. I could not fully verify that only the empty-sample-cases get stuck, but my test runs were consistent with that.