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
298 stars 139 forks source link

Fatal Error. Aborting at EinsplineSetBuilder::AnalyzeTwists2 #2768

Closed kryczko closed 4 years ago

kryczko commented 4 years ago

Describe the bug After running PWSCF and a Jastrow optimization for a 2 atom graphene system, DMC fails with the following error:

Fatal Error. Aborting at EinsplineSetBuilder::AnalyzeTwists2

This error appears in version 3.9.2 and the latest version of the code.

To Reproduce The link to the files which reproduced this error is here: https://drive.google.com/file/d/1bk_h4TWQHhKgN6KO9r3AHRfuH8fBWfle/view?usp=sharing

Expected behavior DMC to run as expected.

System: Cori-haswell.

Additional context Add any other context about the problem here.

prckent commented 4 years ago

@jtkrogel Are you able to see if this is a NEXUS related error or a QMCPACK error? Presumably some has gone wrong with bundling of the twists.

jtkrogel commented 4 years ago

I've actually given this a pretty close look. There are 16 twists in the h5 file, each twist is requested by index (0-15) in the xml input (little chance for input error), each twist is apparently found by QMCPACK, but then 12 of the 16 twists abort in AnalyzeTwists2 (4 run past the twist analysis).

I think this is a bug in the AnalyzeTwists2 routine in QMCPACK.

prckent commented 4 years ago

It would be interesting to know if the problem occurs without any bundling. i.e. This is definitely some problem with the twist analysis.

kryczko commented 4 years ago

I found the root cause after running just 1 twist as suggested by @jtkrogel. The error was:

ERROR Cannot use this super twist with real wavefunctions.
Please recompile with QMC_COMPLEX=1.

I will recompile qmcpack with this option and report back.

kryczko commented 4 years ago

I am now getting the error:

Fatal Error. Aborting at ParticleSet::addTable Cannot mix AoS and SoA distance tables.

I ran a single twist and I receive the same error.

kryczko commented 4 years ago

this issue seems to be fixed for commit id: 035e99ecd4047a805ca86326257ee3509f316031

prckent commented 4 years ago

Thanks for chasing this. Q. That version and older is OK for you?

kryczko commented 4 years ago

Yes this version is ok. I can also double check the latest version.

Kevin

On Oct 23, 2020, at 12:17 PM, Paul R. C. Kent notifications@github.com wrote:

 That version and older is OK for you?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.

ye-luo commented 4 years ago

this issue seems to be fixed for commit id: 035e99e

This commit itself is unlikely a fix to your issue but probably something before it.

jtkrogel commented 4 years ago

If you are running with the energy density on, you might run into this distance table error for v3.9.2. All the energy density specific updates were made after this point. Your original issue appears to be limited to real vs. complex code.

kryczko commented 4 years ago

yes, the real vs. complex code was the issue. Going to close.