-
Notifications
You must be signed in to change notification settings - Fork 9
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
downloader.py sets method and basis non-deterministically #44
Comments
I couldn't reproduce the error, I see the zeroth element is always I ran
and the output shows:
@leifjacobson Can you please drop the version of qcportal you are using. cc: @bennybp |
|
The downloader script is versioned as part of the repository. Even if we change it for future versions, someone who wants v1.1 of the dataset is going to download or check out that version of the repository and run the corresponding version of the script. So we need to make sure it works correctly with the existing script. |
Seems consistent within invocations but can change between them?
|
Thank you for the report @leifjacobson. This is in @bennybp 's territory, I will let him know on slack.
We should make that version inaccessible and make a minor release then (v1.1.1), unfortunately I don't see any other way. @dotsdl or @jthorton can you take a look and confirm when you can. |
@peastman : If the download script does not hard-code the
You can erase the tag and re-release it, but I'd suggest a simple 1.1.1 bugfix with a note in the release notes that this eliminates the non-determinism error would be sufficient. |
Can you make a PR with the fix? I'm not sure exactly what it is. |
The fix is merged and the 1.1.1 release is up. Thanks @pavankum for the very fast response. |
I'm was trying to reproduce the energies in a few examples of your SPICE dataset as downloaded with
downloader.py
. I was unable to successfully reproduce the energies and after some debugging discovered thatspec['method']
andspec['basis']
was set tob3lyp
andtzvp
respectively. I further found that repeated invocations ofdownloader.py
can result in setting these values in a non-deterministic way. Since these values are retrieved inside a loop over subsets I suspect this could also result in a mixed dataset.I'm downloading again after hardcoding the appropriate values and will try to reproduce the energies again. Regardless of outcome I suggest this is investigated and fixed.
The text was updated successfully, but these errors were encountered: