Fix device OneDFT gradient#215
Open
tvogels wants to merge 1 commit into
Open
Conversation
The OneDFT GPU gradients gave incorrect results due to a mismatch in task ordering. GauXC splits grid points in batches (tasks). The OneDFT code path sorts these by atom, because the neural network treats atomic grids as units that belong together. When computing nuclear gradients, we use pytorch to compute gradients w.r.t. model input features (density, kinetic energy density, ...) and combine them with gradients of those input features w.r.t. the coordinates. When those are combined, the ordering still needs to match up. So this PR restores correct ordering + adds a test comparing host gradients to device gradients. This test failed before this PR.
jhrmnn
approved these changes
Jun 26, 2026
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
The OneDFT GPU gradients gave incorrect results due to a mismatch in task ordering.
GauXC splits grid points in batches (tasks). The OneDFT code path sorts these by atom, because the neural network treats atomic grids as units that belong together.
When computing nuclear gradients, we use pytorch to compute gradients w.r.t. model input features (density, kinetic energy density, ...) and combine them with gradients of those input features w.r.t. the coordinates. When those are combined, the ordering still needs to match up.
So this PR restores correct ordering + adds a test comparing host gradients to device gradients. This test failed before this PR.