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
41 stars 54 forks source link

Update MASTODON install instructions (using conda to activate MOOSE) #307

Closed somu15 closed 4 years ago

somu15 commented 4 years ago

With the conda activate moose and conda update --all commands, libmesh need not be manually built or updated. Please see instructions here on the MOOSE website. This issue can make MASTODON up-to-date with the MOOSE developments.

cbolisetti commented 4 years ago

Yes, this has existed for a while but some of our tests will fail if we use the conda package because the conda package uses more recent compilers. When I tried this several months ago, quite a few tests failed, but @hoffwm mentioned that only 1-2 tests fail now. We might only have to regold these. @hoffwm since you are already on conda, can you verify this and regold them? @somu15 can then update the instructions.

lynnmunday commented 4 years ago

I put in similar instructions for building blackbear with conda since it needs the libmesh version of boost like mastadon. Why does mastadon need the full version of boost from libmesh? Ben changed the stochastic_tools distributions over to not use boost that we might be able to call directly. Would that work? Blackbear needed boost for other reasons.

https://mooseframework.org/blackbear/source/materials/NEMLStress.html

lynnmunday commented 4 years ago

@cbolisetti @hoffwm I can run the tests, I'm using conda. Are they all regression tests?

lynnmunday commented 4 years ago

test:materials/isoil.data_file .............................................................. FAILED (EXODIFF) test:bcs/seismic_force.seismic_force_test_z ................................................. FAILED (EXODIFF)

These two fail You want them reblessed?

cbolisetti commented 4 years ago

@lynnmunday thanks for running those! Yes, please rebless them if the differences are really small. If not, we'll have to look into it in more detail. Can you post the exodiff results here?

cbolisetti commented 4 years ago

Why does mastadon need the full version of boost from libmesh? Ben changed the stochastic_tools distributions over to not use boost that we might be able to call directly. Would that work? Blackbear needed boost for other reasons.

We need to remove the BOOST dependency -- that's on my list of things to do..

lynnmunday commented 4 years ago

We need to remove the BOOST dependency -- that's on my list of things to do.. Do you have an issue for this, I can do it.

lynnmunday commented 4 years ago

test:materials/isoil.data_file: --------- Time step 782, 7.8100000e+00 ~ 7.8100000e+00, rel diff: 0.00000e+00 --------- test:materials/isoil.data_file: Global variables: test:materials/isoil.data_file: Nodal variables: test:materials/isoil.data_file: Element variables: test:materials/isoil.data_file: --------- Time step 783, 7.8200000e+00 ~ 7.8200000e+00, rel diff: 0.00000e+00 --------- test:materials/isoil.data_file: Global variables: test:materials/isoil.data_file: accel_6z rel diff: 3.4966884e-08 ~ 3.4965971e-08 = 2.61060e-05 (FAILED) test:materials/isoil.data_file: Nodal variables: test:materials/isoil.data_file: accel_z rel diff: 3.4966884e-08 ~ 3.4965971e-08 = 2.61060e-05 (node 6)

test:bcs/seismic_force.seismic_force_test_z: --------- Time step 177, 1.7600000e+00 ~ 1.7600000e+00, rel diff: 0.00000e+00 --------- test:bcs/seismic_force.seismic_force_test_z: Global variables: test:bcs/seismic_force.seismic_force_test_z: Nodal variables: test:bcs/seismic_force.seismic_force_test_z: Element variables: test:bcs/seismic_force.seismic_force_test_z: --------- Time step 178, 1.7700000e+00 ~ 1.7700000e+00, rel diff: 0.00000e+00 --------- test:bcs/seismic_force.seismic_force_test_z: Global variables: test:bcs/seismic_force.seismic_force_test_z: Nodal variables: test:bcs/seismic_force.seismic_force_test_z: accel_z rel diff: -1.2179614e-03 ~ -1.2179341e-03 = 2.23994e-05 (node 17) test:bcs/seismic_force.seismic_force_test_z: Element variables:

lynnmunday commented 4 years ago

@lynnmunday thanks for running those! Yes, please rebless them if the differences are really small. If not, we'll have to look into it in more detail. Can you post the exodiff results here?

I pasted the exodiff above.
For seismic force, 1e-3 is really small, the max for accel_z is 206.55778 For isoil.data_file. accel_6z is 1e-1. and disp_6z is 8e-5

cbolisetti commented 4 years ago

Yeah, those look really small -- can you please regold and create a PR? I wanted to do it earlier but converting back and forth from conda to the MOOSE package takes a lot of time..