Clarification question about conj_ht test

[aprilnovak]

For the conj_ht CI test, what is the significance of the numbers used for checking the solution? Is there an analytic solution for this flow, or were these just numbers selected at the time the test was created based on a “reasonable” looking solution? Asking in order to support SQA NEAMS milestones.

NekRS checks that the following are satisfied:

• Temperature at (x,y,z) = (4, 4, 4) is 13.921
• Temperature at (x,y,z) = (-0.25, 0.25, 1) is 0.00133
• Temperature at (x,y,z) = (0.5, 0.5, 0.5) is 9.62969

There are some checks on norms of velocity and temperature as well. Where do these numbers (2.06559 and 5.16746) come from?

  vxErr = abs((err[0] - 2.06559) / err[0]);
  sErr = abs((err[1] - 5.16746) / err[1]);

Thank you!

[stgeke]

This is a regression test comparing the solution against nek5000.

…

On Fri, 3 Jan 2025 at 23:44, April Novak ***@***.***> wrote: For the conj_ht CI test, what is the significance of the numbers used for checking the solution? Is there an analytic solution for this flow, or were these just numbers selected at the time the test was created based on a “reasonable” looking solution? Asking in order to support SQA NEAMS milestones. NekRS checks that the following are satisfied: - Temperature at (x,y,z) = (4, 4, 4) is 13.921 - Temperature at (x,y,z) = (-0.25, 0.25, 1) is 0.00133 - Temperature at (x,y,z) = (0.5, 0.5, 0.5) is 9.62969 There are some checks on norms of velocity and temperature as well. Where do these numbers (2.06559 and 5.16746) come from? vxErr = abs((err[0] - 2.06559) / err[0]); sErr = abs((err[1] - 5.16746) / err[1]); Thank you! — Reply to this email directly, view it on GitHub <#613>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AEN2M743ACDWQL27R2677M32I4HEFAVCNFSM6AAAAABUSOJOZGVHI2DSMVQWIX3LMV43ERDJONRXK43TNFXW4OZXG43TKOBTGA> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[aprilnovak]

Thank you. Is the Nek5000 solution itself verified against anything? We require this type of explainability for the QA process, since Nek5000 has not gone through a similar QA release process.

If not, that is ok -- we'd just interpret this test to be a "looks reasonable, let's stamp this solution as good and make sure future modifications don't change any behavior."