search

QMCPACK / qmcpack

Main repository for QMCPACK, an open-source production level many-body ab initio Quantum Monte Carlo code for computing the electronic structure of atoms, molecules, and solids with full performance portable GPU support
http://www.qmcpack.org
Other
297 stars 139 forks source link

Tests wanted / List of untested functionality #1061

Open jtkrogel opened 6 years ago

jtkrogel commented 6 years ago

Many tests have been added, but there still remain many features that need to have tests added. The list below is meant to capture all features that do not yet have a test, or known breakages that do not yet have a corresponding test so that fixes can be pursued and validated.

Real space statistical integration tests needed

Real space parallelization tests needed:

Real space estimator tests needed:

Real space wavefunction tests needed:

Real space method tests needed:

Tests that might be useful for bugfixes:

Real space deterministic integration tests needed

(plan out first set of tests to add here, see also #704 )

Real space unit tests needed

(need to list)

Auxiliary field integration tests needed

Auxiliary field estimator tests needed:

Auxiliary field unit tests needed:

Auxiliary field method tests needed:

See #1062 for an overarching view of the testing problem and links to its other aspects.

jtkrogel commented 6 years ago

@rcclay are there other RMC features that aren't covered yet?

@mmorale3 should this be filled out for AFQMC now, or following the mainline merge of your updated AFQMC?

rcclay commented 6 years ago

@jtkrogel Not that I can think of off the top of my head.

Hyeondeok-Shin commented 5 years ago

I'm now working with the real space wavefunction tests (including deterministic test), but please let me know if someone is already working with a certain test in order to avoid conflict.

jtkrogel commented 5 years ago

@Hyeondeok-Shin You mean via the wavefunction tester? I expect pass/fail criteria for the finite differences will require some tinkering. If this is the case, perhaps make a new issue for further discussion on that topic.

Please can you add the deterministic integration tests you will be working on next to the header comment under the appropriate section?

Hyeondeok-Shin commented 5 years ago

@jtkrogel I see. I'll be adding the deterministic integration tests list I'm (will be) working on.

prckent commented 5 years ago

Please steer clear of the "wavefunction tester" for new tests. It is another layer of indirection and uncertainty to investigate when something goes wrong.

A test of traces would be preferred by far, be simpler, and achieve the same coverage.

prckent commented 4 years ago

Renamed for clarity.

walshmm commented 3 years ago

Was wondering how up to date this issue is. Wanted to look at where I could start writing tests, though I dont want to step on anyone's toes as I understand some parts of the code are being refactored.

prckent commented 3 years ago

It is fairly up to date. Since this indexing issue was created we have added use of codecov, which also is a good guide to where tests might be missing. Of course the reported codecov coverage https://app.codecov.io/gh/QMCPACK/qmcpack only sees the tests run in CI, not our complete set of tests, so the coverage is somewhat better.

Many of these require either wholly new input files to use the functionality or they require a bit of infrastructure. e.g. https://github.com/QMCPACK/qmcpack/issues/1103 is important to test since it is a production feature, but new cmake macros would be needed to copy multiple input files, run "qmcpack listofinputs.txt", and then analyze the results. The individual runs can be repeats of runs where we have existing tests.