travs
commented
5 years ago Ok so have fixed points 1 and 2. For 2, the distribution histograms retain outliers so the plots may look "zoomed out". We can adjust this if we wish, but they still show that the distributions are very tight.
For the notebooks, it might be helpful if we establish the convention of removing output cells as we develop a notebook. These cells don't play well with version control, because they are generated and binary, so it makes diffs harder to understand, and bloats the repo in the long term.
For now we can use Cell -> All output -> Clear in the notebook interface before we commit.
If this gets annoying and we want to automate it we can look at these solutions or these tips.
1) For some reason, the way I've implemented the dynamic updating of the simulation trajectory, after the run is complete, it outputs the entire trajectory frame by frame (you can see this in the notebook). This needs to be fixed.
2) Adding a small function to generate many runs of the simulation and plot histograms of the parameters so that we can visualize the robustness of simulation results (see the pinned plot in the Slack channel).
3) Any other generic Python improvements would be appreciated.
@travs