Segmentation fault

[AnsgarSiemer]

Thank you for RotamerConvolveMD the program is very helpful to our research.

I have problems with segmentation faults for some distance pairs of a MD simulation. The program runs for a while and then crashes at a certain frame.

Step 1057/4442 [ 23.8%]Segmentation fault<

Increasing the clashDistance changes the frame the segmentation fault occurs. If I start the program a few frames before the frame where the segmentation fault occurred (e.g. by setting --discard=1055), I cannot reproduce the segmentation fault (i.e. I cannot really narrow it down to a specific frame). The test scripts work without any problem. I tried this on Debian buster and bullseye both using native python (2.7 and 3.7) and anaconda as well as anaconda on an OSX system and got the same error.

Any suggestions what I could do to solve or narrow down the problem?

[orbeckst]

This sounds like a difficult bug to debug. At which line in the Python code does the segfault occur?

What version of MDAnalysis do you have installed?

import MDAnalysis
print(MDAnalysis.__version__)

Perhaps it's a bug in the distance calculation code??

That said, we're not actively maintaining RotamerConvolve anymore but we're more than happy to accept fixes into the code base.

Also note that we're only loosely monitoring this issue tracker but you're welcome to talk t us on the MDANalysis user mailing list https://groups.google.com/forum/#!forum/mdnalysis-discussion