NU Refinement continues from last failed iteration

Dear CryoSPARC developers,

Is it possible to have a new feature to allow NU Refinement to continue from the last failed iteration? I had a NU Refinement job running on a cluster which has a strict job running time limit, but my NU Refinement job could not be done within the time limit. Thanks!

3 Likes

I would like to second this request: we’ve had some network instability that would cause jobs to fail ~24 hours in, so not having to start from scratch would be nice.

2 Likes

Thank you for this suggestion. This feature is unfortunately not on our near-term roadmap.
You may shorten the runtime by down-sampling the particles, trading speedier calculations against some loss in information. The optimal balance of course depends on the data and the particular modeling problem.
Also, would the manager(s) of your cluster be open to accommodating longer runtimes, if presented with a strong case?

No. Our cluster administrator would not do that unfortunately although I requested several times.