Ab-initio error: ValueError: could not broadcast input array from shape (860875,) into shape (100000,)

3D Reconstruction
1

Hi,

I want to run ab-initio without using all of my particles. So my input parameters in Number of Ab-Initio classes:8; Num particles to use:100000. But sometimes there was an error as follows:

image
image1120×137 21.1 KB

It not happened every time. My cryoSparc version is : v4.5.3
If I use all particles, let the Num particles to use as empty. It finished successfully.

Do you have an idea about that?

Best,
Wendy

2

@Wendy Please can you

  • post approx 20 lines from the Event Log that immediately precede the Traceback
  • let us know whether
    • particle caching was used for the failed job
    • instances of the job may variably fail or succeed even with same particle inputs
    • instances of the job have always failed after the first failure
3

Hi wtempel,

Here is all Event Log:

image
image1120×697 95.7 KB

I use the same particles to do ab-inito, and just the change particle numbers in the parameter of Num particles to use in ab-initio as using all particles cost a lot of time. If I set Num particles as 100000 or 200000, it fails, while I set it empty, It success.

Thanks,
Wendy

4

This error may occur in ab-initio reconstruction jobs in CryoSPARC versions 4.5.2 and 4.5.3 where:

  • particles have already been assigned to half-sets and
  • Num particles to use has been specified

As a workaround one can split the input particles using a Particle Sets Tool job to create a particle set of the desired size for input to ab initio 3D reconstruction. For downstream refinement, one may connect the volumes from ab initio reconstruction along with the full particle set.

1 Like
5

Got it .
Thanks, wtempel.

Best,
Wendy