search

idaholab / mastodon

A MOOSE app for structural dynamics, seismic analysis, and risk assessment.
https://mooseframework.inl.gov/mastodon
GNU Lesser General Public License v2.1
40 stars 52 forks source link

Fixing test tolerance #408

Closed somu15 closed 2 years ago

somu15 commented 2 years ago

Fixing test tolerance closes #408

somu15 commented 2 years ago

@cbolisetti Can you please merge this if the tests pass? @fdkong needs to update libmesh and this test is causing tolerance issues. Thanks!

fdkong commented 2 years ago

@somu15 Are you able to make the solver converge more accurate? For example, make the solver relative tolerance smaller.

I will check this against new PETSc before we merge.

somu15 commented 2 years ago

@fdkong I increased the abs_tol and tightened the rel_tol. Is this want you want?

fdkong commented 2 years ago

I would like to make the test more robust ...

I am checking now

fdkong commented 2 years ago

@somu15 Do you know why we have mesh in this way?

[Mesh]
  type = GeneratedMesh
  dim = 3
  nx = 1
  ny = 1
  nz = 20
  xmin = 0.0
  xmax = 25.0
  ymin = 0.0
  ymax = 25.0
  zmin = 0.0
  zmax = 5000.0
[]

In other words, we do not have really have any interior nodes, and what are we solving for? Everything is on boundary

somu15 commented 2 years ago

@fdkong May be there are interior nodes in the z direction? The test description says the mesh has 20 elements in the z direction to test a non-reflecting BC for seismic waves. It only has 20 elements for testing purposes.

moosebuild commented 2 years ago

Job Test on 220ec36 wanted to post the following:

View the site here

This comment will be updated on new commits.

fdkong commented 2 years ago

May be there are interior nodes in the z direction?

There are no interior nodes. Apply PBCs generate some unknowns

moosebuild commented 2 years ago

Job Coverage on 220ec36 wanted to post the following:

Coverage

Coverage did not change

Full coverage report

This comment will be updated on new commits.

fdkong commented 2 years ago

@lindsayad

Could you take a look at this PR? I tried to refine the mesh so that the solution is more accurate and more stable.

The gold file was generated using PETSc-3.16, and it could pass the tests of PETSc-3.11 and PETSc-3.15

I think we are ready to merge

@somu15 Do you have any comment ?