This issue also just popped up for me. However, my NU Refine job is running just fine. We’re running v4.6.2, as well.
@jrgib Did you already try disabling Minimize over per-particle scale during local refinement, as reported in Local refinement error (cryoSPARC v4.5.3) - #12 by lazaret?
The job did fail with the minimize over per-particle scale value turned on. After running a NU refinement (without the minimize over per-particle scale on), I ran another local refinement job using the inputs from the NU refinement, and the job ran just fine. I haven’t run into this error since then.
I encountered the same error in version 4.7.1.
In my case, I needed to run a local refinement job with per-particle scale optimization as preparation for a particle subtraction job. The input reference and mask had been shifted (re-centered) using the Volume Align Tools job to focus on a peripheral region of my target protein. The particles were also shifted in the same volume alignment job and then re-extracted from the micrographs.
I tried several different parameter settings for local refinement, and what worked for me was changing “Scale min/use start iter” from the default value of 0 to 2.
Custom parameters I used:
- Use pose/shift gaussian prior: ON
- Stdev of pose/shift prior: 1, 1
- Re-estimate greyscale level of input reference: ON
- Reset input per-particle scale: ON
- Minimize over per-particle scale: ON
- Scale min/use start iter: 2
I’m not sure if changing the start iteration will generally work, but I hope this information helps.