NU refinement in v5.0.2

3D Refinement
1

After upgrading to cryoSPARC v5, NU refinement completes successfully with the settings listed below:

image
image849×461 16.1 KB

However, when switching to the following configuration:

image
image844×458 16.1 KB

the job crashes near the final iteration and returns the following error:

> Feb 26, 2026, 7:16:59 PM

Traceback (most recent call last):
File “cli/run.py”, line 106, in cli.run.run_job
File “cli/run.py”, line 211, in cli.run.run_job_function
File “compute/jobs/refine/run.py”, line 604, in compute.jobs.refine.run.run_homo_refine
File “compute/jobs/refine/run.py”, line 605, in compute.jobs.refine.run.run_homo_refine
File “compute/jobs/ctf_refinement/run.py”, line 436, in compute.jobs.ctf_refinement.run.full_ctf_refine
File “/home/spuser/Downloads/cryosparc/cryosparc_worker/.pixi/envs/worker/lib/python3.12/site-packages/numpy/linalg/linalg.py”, line 409, in solve
r = gufunc(a, b, signature=signature, extobj=extobj)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File “/home/spuser/Downloads/cryosparc/cryosparc_worker/.pixi/envs/worker/lib/python3.12/site-packages/numpy/linalg/linalg.py”, line 112, in _raise_linalgerror_singular
raise LinAlgError(“Singular matrix”)
numpy.linalg.LinAlgError: Singular matrix

How can I solve this problem?

2

I would recommend not running all Global CTF refinements during a 3D refinement run (whether homogeneous or non-uniform).

Check through previous iterations to see whether spherical aberration or tetrafoil have been drifting off a long way.

If you only enable tetrafoil, or aniso mag, does it complete successfully?

3

Thank you for your response. These settings were working properly before the upgrade, and the errors only started occurring after updating. I’m not sure what might be causing this issue.

Additionally, I tried turning on anisotropic magnification refinement alone, and the job completed successfully. However, the resolution of the output map was noticeably worse.

4

I’m curious about Global CTF parameters, although it may not be related.

Also check dmesg as soon as the error hits to see whether there are any GPU related errors.

5

Hi @Songdf ,

Are you able to share the full job reports from all of:

  • The previously successful (pre-v5) NU refine
  • The v5 NU refine that crashed
  • The job in which you:

    turned on anisotropic magnification refinement alone

Best,
Michael