Local refinement fails at iteration 0 Running local cross validation for A


I have been trying to run local refinement to mask out one subunit of an icosahedral virus. However, the jobs always fail at the Running local cross validation for A … step for iteration 0. The job only gives the error job process terminated abnormally.

This happens for both full particles and subtracted particles which have been symmetry expanded following a homogeneous refinement job.

The parameters of the job have been left as default.

I have no idea about where to start to fix this issue, so any ideas are appreciated.

What GPU do you have? I’ve noticed this as well an RTX 2080, but re-launching on a machine with an RTX 2080 Ti successfully completes the refinement. I was wondering if it was a cryptic memory issue or something

We have a RTX 3080 ti. However, the box size is large 768. I will try downsampling the particles and see if this helps (if it is indeed a memory issue).

Both downsampling and reducing the number of particles has not helped.

Please can you post job.json and job.log from the failed job’s directory.
What are the

  • reduced box size
  • number of particles


It eventually worked when I reduced the particle number further. Now I have 700,000 particles and a box size of 384