search

JohannesBuchner / UltraNest

Fit and compare complex models reliably and rapidly. Advanced nested sampling.
https://johannesbuchner.github.io/UltraNest/
Other
142 stars 30 forks source link

post_summary.csv files generated by ultranest logging have out of order data values #78

Open jvshields opened 1 year ago

jvshields commented 1 year ago

logging post_summary.csv files generated by ultranest when specifying a logging directory in ReactiveNestedSampler have values different from what what the column names suggest.

The ultranest documentation suggestions that you can ingest these files with pandas.read_csv() to get out the summary statistics for the fitted parameters. These files have column names along the lines of: param_1_mean, param_1_stdev ... param_2_mean, param_2_stdev...

Instead, it looks like the data values are organized as: param_1_mean, param_2_mean ... param_1_stdev, param_2_stdev...

so when you ingest them you get mismatches that can't be used for looking at the summary statistics of the output parameter distributions.

JohannesBuchner commented 1 year ago

Please check whether this is still a problem in the latest version (3.5.7).

JohannesBuchner commented 1 year ago

I think this was fixed in v3.4.6 by commit 58466d40024d2144c50d743913037d1813862e72

jvshields commented 1 year ago

Sorry, I would try this if I could easily do so, but conda informs me that other dependencies of my fitting routine are incompatible with ultranest beyond 3.3.0.