Skip to content

Lower the threshold for tracer1 convergence in rotation_2d#488

Closed
xylar wants to merge 1 commit intoE3SM-Project:mainfrom
xylar:update-rotation-2d-thresholds
Closed

Lower the threshold for tracer1 convergence in rotation_2d#488
xylar wants to merge 1 commit intoE3SM-Project:mainfrom
xylar:update-rotation-2d-thresholds

Conversation

@xylar
Copy link
Copy Markdown
Collaborator

@xylar xylar commented Mar 11, 2026

This is necessary on Chrysalis, where JIGSAW seems to be producing a mesh at QU90km resolution that results in an unusually noisy tracer1 field. In general, QU meshes result in noise tracer1 that does not decrease monotonically with increasing resolution over the range of resolutions we are testing. See #487 for details.

Checklist

  • Testing comment in the PR documents testing used to verify the changes

fixes #487

@xylar
Copy link
Copy Markdown
Collaborator Author

xylar commented Mar 11, 2026

Testing

I tested this in conjunction with the submodule update in #485. Together, these allow all omega_pr tests to pass on Chrysalis with Intel.

@xylar xylar requested a review from cbegeman March 11, 2026 09:43
@xylar xylar self-assigned this Mar 11, 2026
@xylar
Copy link
Copy Markdown
Collaborator Author

xylar commented Mar 11, 2026

@cbegeman, can you let me know if you're okay with this (based on my plots in #487)?

@xylar
Copy link
Copy Markdown
Collaborator Author

xylar commented Mar 11, 2026

Aaaargh! I don't know what's going on. I ran omega_pr to test #489 without these changes and it ran fine:

Order of convergence for tracer1: 0.712
Order of convergence for tracer2: 1.576
Order of convergence for tracer3: 0.331

I don't know why I'm not getting consistent results.

This is necessary on Chrysalis, where JIGSAW seems to be
producing a mesh at QU90km resolution that results in an unusually
noisy tracer1 field.  In general, QU meshes result in noise
tracer1 that does not decrease monotonically with increasing
resolution over the range of resolutions we are testing.
@xylar xylar force-pushed the update-rotation-2d-thresholds branch from 1d1f9b5 to 1a3f227 Compare March 11, 2026 14:16
@xylar
Copy link
Copy Markdown
Collaborator Author

xylar commented Mar 11, 2026

I did a clean checkout of everything, and rebuilt jigsaw from scratch. Things are now working for me. I don't think this PR is needed after all.

@xylar xylar closed this Mar 11, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

qu/rotation_2d tracer 1 convergence rate much slower after switch to PCD

1 participant