Refactor and unify the random numbers.

[cianciosa]

This unifies and cleans up the random number generation to ensure that random numbers are seeded once at the beginning.

[quantumsteve]

@idigs @cianciosa Do you see an error message from job_two?

[cianciosa]

@idigs @cianciosa Do you see an error message from job_two?

I'm guessing the test failure is a result of differences in seeding. Do you know if this test uses a constant seed?

[mbeidler3]

The initial case I provided may use a different seed than the present case.

[quantumsteve]

@idigs @cianciosa Do you see an error message from job_two?

I'm guessing the test failure is a result of differences in seeding. Do you know if this test uses a constant seed?

I think it's a Chrome or GitHub issue.

[quantumsteve]

[cianciosa]

I get

[mbeidler3]

This is looking pretty good on my computer. When I run it multiple times, I keep getting the same exact results with a good distributions of REs. The only issue is that I am getting different answers compared to what Mark sent me by email on 6/5/24. While the spatial distribution is different.

[mbeidler3]

test this please

[mbeidler3]

@idigs it looks like the SPACK cache is getting built and created again. Can we set it up a workflow to have this to happen automatically on a weekend evening?

[mbeidler3]

test this please

[mbeidler3]

@idigs, how long does it usually take for an ExCL runner to pick up the Onyx workflow?

[mbeidler3]

Didn't do comprehensive testing due to size, but works for mars regression CTest.