search

LLNL / libROM

Model reduction library with an emphasis on large scale parallelism and linear subspace methods
https://www.librom.net
Other
201 stars 36 forks source link

Regression Tests(addition to the PR: try compiling mfem with gitlab actions) #124

Closed kevinhkhuynh closed 1 year ago

kevinhkhuynh commented 2 years ago

The goal here is to compile libROM with MFEM for Gitlab actions so that we can run the example scripts as part of a test suite.

chldkdtn commented 2 years ago

@kevinhkhuynh do you think you need to pass this to @pranav-vempati as well?

kevinhkhuynh commented 2 years ago

For some reason, for Linux, the test is failing at the example run due to "Operating system error: Cannot allocate memory"

kevinhkhuynh commented 2 years ago

For some reason, for Linux, the test is failing at the example run due to "Operating system error: Cannot allocate memory"

@axla-io Axel, didn't you run into a similar issue on your virtual machine, relating to the cannot allocate memory? How did you end up fixing it?

axla-io commented 2 years ago

For some reason, for Linux, the test is failing at the example run due to "Operating system error: Cannot allocate memory"

@axla-io Axel, didn't you run into a similar issue on your virtual machine, relating to the cannot allocate memory? How did you end up fixing it?

Yes I did! In the end I allocated more memory and more swap memory for my vm. Did you set the mc-model = large flag also?

chldkdtn commented 1 year ago

@dylan-copeland @siuwuncheung Pranav needs your help on reviewing this PR. The PR is not complete yet, but it would be very helpful for you to start reviewing and point out what is missing to Pranav. Then he can fill the gap.

dylan-copeland commented 1 year ago

@pranav-vempati Should regression_tests/simulationLogFile be in this PR?

pranav-vempati commented 1 year ago

All suggestions except the comparison of examples' relative errors are now incorporated in this PR.

chldkdtn commented 1 year ago

@dylan-copeland @siuwuncheung , @pranav-vempati has just informed me that this PR is ready for review except continuous integration part. All the regression tests must be working locally, so we can start reviewing the local regression test implementation. I suggest not to wait until the CI is fixed. Let's start giving @pranav-vempati valuable feedbacks on this PR.

chldkdtn commented 1 year ago

@pranav-vempati, I intentionally broke one of dmd examples and ran regression tests. The regression tests correctly identified the failing example, but I cannot find any log or output files which tells me what exactly broke. If there is such an output file, can you tell me where? If not, can you modify the regression test so that it outputs the error or the reason for failure?

chldkdtn commented 1 year ago

@pranav-vempati, can you add instruction of how to add a new regression test?

pranav-vempati commented 1 year ago

@chldkdtn Instructions on how to add a new regression test have been added to REGRESSIONTEST.md.

pranav-vempati commented 1 year ago

Thanks @pranav-vempati! The tests all pass on linux and mac. One small suggestion is that the squash-and-merge has resulted in some very long commit messages in the libROM history. Since this PR has almost 200 commits, it would be a good idea to replace all those commit messages with a brief summary of this PR.

@dylan-copeland Thanks for the suggestion. I will do this before I merge this PR.