search

MALBECC / lio

Linear implementation of DFT calculations (CPU and GPU)
GNU General Public License v2.0
26 stars 18 forks source link

Testamber broken #197

Closed ramirezfranciscof closed 6 years ago

ramirezfranciscof commented 6 years ago

So, apparently something broke in testamber somewhere in the last months. I'm not sure how much of this is specific to the test or how much is a more general amber interface breakdown. It might be as stupid as a keyword lack of update, but I'll have to check it out.

0fb304d1f5a6fdb9951729f0dc468aeb0b383555 - Working ("Ehrenfest now supports multiple time steps.") f15bd36b52836645220eadd2ae7eeb91aaf19542 - Wrong numbers ("Ehrenfest PR02 - Incorporation of multiple time-step scheme.") RC HEAD - Directly broken.

ramirezfranciscof commented 6 years ago

Atompot PR0 still works

fedepedron commented 6 years ago

Good thing the test is broken and not the main code :stuck_out_tongue:

ramirezfranciscof commented 6 years ago

For the record: 06c22a2b18de732c568312201d4a15e4bc8fc5f7 (Pop analysis) => Good numbers 339f549f1bf74019983c79c00101392c73d11418 (DFTB pr 0) => Wrong numbers

Now to check where's the problem, I might eventually need @charlyqchm assistance.

fedepedron commented 6 years ago

This is actually worse than we thought, forces calculation are not working propperly. NVM, this is actually fixed in current RC.

ramirezfranciscof commented 6 years ago

Yes indeed.