Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Different energies and structures for same input with fixed atoms #1164

Closed
markus-hermann opened this issue Jan 31, 2025 · 3 comments
Closed
Labels
unconfirmed This report has not yet been confirmed by the developers

Comments

@markus-hermann
Copy link

markus-hermann commented Jan 31, 2025

Describe the bug
If I run the same optimization with fixed atoms several times, I do not get the same results.
I want to calculate a torsion profile.
For this I rotate a bond with RDKit in 10° steps and give the calculations to xtb.
I fix the atoms that have been rotated and run a structure optimization.
In the following picture is the results of 7 runs of the same molecule with the same inputs.
Often, these results match, but sometimes they don't.
Image

To Reproduce
Steps to reproduce the behaviour:

  1. inputs are in the attached file
  2. xtb input.xyz --input xtb.inp --opt --chrg 0 --uhf 0 --acc 1 --iterations 150 --gfn 2 -P 1
    You sometimes need 10-40 runs for the error to occure. Somehow more likely on an intel node than on AMD.

Expected behaviour
Consistent results.

Additional context
xtb version 6.7.1 (edcfbbe) compiled by 'conda@3ef47c1a292c' on 2024-11-25

@markus-hermann markus-hermann added the unconfirmed This report has not yet been confirmed by the developers label Jan 31, 2025
@toxtran
Copy link

toxtran commented Jan 31, 2025

Fixed by #1125

@markus-hermann
Copy link
Author

markus-hermann commented Feb 3, 2025

Thanks @toxtran,
I changed to 6.6.1 and ran it again (picolinamide_m1_9).
Much better results.
Image
When will 6.7.2 be available in conda-forge?

@marvinfriede
Copy link
Member

We are working on the 6.7.2 release. Hopefully, we will manage to finish everything until the end of February.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
unconfirmed This report has not yet been confirmed by the developers
Projects
None yet
Development

No branches or pull requests

3 participants