Fix vector indexing error in psi::IntegralTransform::process_spaces #1220
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
jturney
approved these changes
Sep 7, 2018
robertodr
approved these changes
Sep 13, 2018
dgasmith
approved these changes
Sep 13, 2018
andysim
reviewed
Sep 13, 2018
| @@ -275,7 +275,7 @@ void IntegralTransform::process_spaces() { | |||
| if (mosym_[orb] == h) { | |||
| aOrbsPI[h]++; | |||
| aOrbSym[count] = h; | |||
| if (aIndex) aIndex[count] = aindex[n]; | |||
| //if (aIndex) aIndex[count] = aindex[n]; | |||
There was a problem hiding this comment.
I think I'm being slow here, but I don't see what the problem is. The (poorly named) aIndex pointer is created if custom array indexing was provided via aindex. If that vector is empty, no copy is attempted; if not, the loops should essentially reorder the aindex vector into aIndex, but sorted by irrep. There should probably be a catch to ensure that the number of orbitals and indices is the same, but I think that exists somewhere upstream.
There was a problem hiding this comment.
Is there a particular compiler error or failed test that led to this change?
There was a problem hiding this comment.
Also, whatever was the reason for this change, why does it not come into play for the beta orbital indexing where the logic is identical with different variable names?
There was a problem hiding this comment.
On Windows, the tests
- casscf-fzc-sp
- casscf-sa-sp
- casscf-semi
- casscf-sp
- rasscf-sp
- molden1
- molden2
crash with run-time vector index error, which happens on that line. In case of beta orbitals, maybe there are no tests for that.
So I not sure, how to make a proper fix.
|
This is also happens on Linux!
|
|
Just on Linux it doesn't cause a crash for a normal build! |
|
This is very helpful - thank you very much. I'll figure out where the problem is soon and send a PR to you. Commenting the line out will fix the problem, but will impact anybody who uses the library with custom MO spaces and the IWL integral format (which, admittedly, should be 0 people). Rather than suddenly deprecate the feature, I'd like to understand where the bad call is happening and fix that instead. Thank you for your patience. |
|
I finally figured out where the problem was, and have created a PR to fix it. Once that PR is merged into your branch, this fix should be good to go. Sorry for taking so long on this - it was quite a tricky issue to figure out. |
Fix DETCI integral transformation setup
andysim
approved these changes
Sep 23, 2018
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Description
This is part of Psi4 porting to Windows (#933).
Todos
Notable points (developer or user-interest) that this PR has or will accomplish.
psi::IntegralTransform::process_spacesQuestions
Checklist
Tests added for any new featuresStatus