dftd3 / simple-dftd3

reimplementation of the DFT-D3 program
https://dftd3.readthedocs.io
GNU Lesser General Public License v3.0
48 stars 20 forks source link

Implementation of Actinides. #48

Closed lukaswittmann closed 8 months ago

lukaswittmann commented 8 months ago

Description

Added actinides support to the Fortran version, following the Torch autodiff DFT-D3 implementation by FreeSoulIG.

Changes Made

Crosschecked and validated results from both implementations to ensure accuracy. The test suite successfully completed without any errors.

awvwgk commented 8 months ago

Thanks for sharing. Looks good to me.

codecov[bot] commented 8 months ago

Codecov Report

Attention: 20 lines in your changes are missing coverage. Please review.

Comparison is base (cdd0a85) 50.15% compared to head (ac3c6d3) 48.63%.

Files Patch % Lines
test/unit/test_model.f90 33.33% 0 Missing and 8 partials :warning:
test/unit/test_dftd3.f90 45.45% 0 Missing and 6 partials :warning:
test/unit/test_ncoord.f90 40.00% 0 Missing and 6 partials :warning:
Additional details and impacted files ```diff @@ Coverage Diff @@ ## main #48 +/- ## ========================================== - Coverage 50.15% 48.63% -1.53% ========================================== Files 34 34 Lines 3776 3707 -69 Branches 1601 1518 -83 ========================================== - Hits 1894 1803 -91 - Misses 1206 1207 +1 - Partials 676 697 +21 ```

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

lukaswittmann commented 8 months ago

Potentially fixed the weird build check error of build (ubuntu-20.04, meson, debug, intel, 2021.2.0) by shortening the lines in the reference file.

lukaswittmann commented 8 months ago

The build (ubuntu-20.04, meson, debug, intel, 2021.2.0) still fails, even after lines were truncated.

I am clueless as to why this is happening. Does anyone else have any insights into the possible reasons for this occurrence?

awvwgk commented 8 months ago

Can't say why the Intel build is failing, there have been issues with the array constructors and number of continuation lines before. It would be mandatory to fix those issues before merging this patch.

lukaswittmann commented 8 months ago

Fixed the issue @awvwgk. The construction of the array was split up into blocks of 2000 parameters; the build finished successfully.