Analising Nested Sampling Chain

[George Kyriacou]

Dear Fabian

I have successfully ran nested sampling within montepython as shown below:

However when trying to analyse through montepython.py info in the NS folder I gain the following errors :

I have also made attempts at plotting the raw files but am not sure on what they specifically are (there are output-.tx , output-ev , output-pos etc) so an understanding of that would also be useful if i am not able to get info to work. I have plotted the output-.tx file and it has similarity to what is in the Kids-450 paper indicating to me that it was not a finnshed run as the error incorrectly suggests.

All help in getting passed this error is much appreciated.

George Kyriacou

[Fabian Köhlinger]

Hi George,

Unfortunately, you ran into another MultiNest bug (related to your previous report). If the path and/or filename for your output are long but not so long so that they would create your previous error message, some of the filenames and/or file-extensions in the NS subfolder are cut off.

If you run montepython info /path/to/your/run/NS/ now on that folder, MontePython will bounce back with the error messages you quote. In principle, this issue should be flagged both in the MontePython and MultiNest repositories but an easy work around is to go to the NS subfolder inside the folder of your current run and rename all corrupted files (technically MontePython only needs a few of those files with correct extensions, so you don't need to rename them all).

Below is a list of the correct file endings of all files the MultiNest run creates inside the NS subfolder:

your_filename.paramnames

your_filename.arguments

your_filename-phys_live.points

your_filename-live.points

your_filename-ev.dat

your_filename-resume.dat

your_filename-summary.txt

your_filename-stats.dat

your_filename-post_equal_weights.dat

your_filename-.txt

After you made sure that all files have the correct ending, just rerun
montepython info /path/to/your/run/NS/ followed by montepython info /path/to/your/run/ to finally get all summary files and plots.

Cheers,

Fabian

[George Kyriacou]

Thank you for the quick response.

I have renamed all files as shown above , unfortunately the exact same error occurs.

[Fabian Köhlinger]

Hi George,

Hmm, mysterious... Could you send me your folder (including the NS subfolder) via Mail? I'll have a look then tomorrow.

Cheers,

Fabian

[George Kyriacou]

Have done so , should be getting an email soon with the subject George's Error , thank you so much.

George