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

Numerical orbital tests are incorrect #2504

Open prckent opened 4 years ago

prckent commented 4 years ago

Describe the bug

https://cdash.qmcpack.org/CDash/testSummary.php?project=1&name=short-c_no-hf_vmc-1-16&date=2020-06-02

Passing tests actually show that the h5 file is not read. An error code should be set. Additionally, several builds are actually crashing.

We may need to discuss the support status of numerical orbitals given the SoA transition.


 Wavefunction setup: 
 ------------------- 
  Name: psi0
  Created SPOSet builder named 'MO' of type mo
  LCAO: SoaAtomicBasisSet<MultiQuintic,1>
  C.s000.h5 version 0 1
  AO BasisSet for C
   Spherical Harmonics  DO NOT contain (-1)^m factor
   Angular momentum m is explicitly given.
   Grid is created by the input paremters in h5
Could not read grid_type in H5; Probably Corrupt H5 file

To Reproduce Steps to reproduce the behavior:

  1. release version or git commit hash being built
  2. cmake command
  3. full program/test invocation command
  4. additional steps

Develop, seemingly any of the "short-c_no" tests.

Expected behavior

Test should only pass when the execution is completed without fault.

System:

All nightly systems - bora, sulfur, nitrogen and with all compilers.

Additional context Add any other context about the problem here.

jtkrogel commented 4 years ago

I think these tests can simply be removed since the SQD atomic HF code was removed some time ago (which is where the orbital file came from) and numerical orbitals were slated to be deprecated with the move to SoA.

prckent commented 4 years ago

My recollection was also that no one spoke up for them. The correct error would then be to abort on encountering the Numerical atomicBasisSet type.

    <atomicBasisSet type="Numerical" elementType="C" expandYlm="no" href="C.s000.h5">
jtkrogel commented 4 years ago

Yes. I can handle this with the residual AoS cleanup.

prckent commented 4 years ago

Thanks Jaron. Worth being alert for other no longer used functionality in the "builder" that handles this.