D-Roberts opened a new pull request #18197: URL: https://github.com/apache/incubator-mxnet/pull/18197
## Description ## This is the 2nd part of the QR backward implementation. The 1st part (merged) covered square and deep matrix shapes (nrows >= ncols) and part 2 now covers the remaining wide matrix shapes (ncols > nrows). References: [Differential Programming Tensor Networks](https://journals.aps.org/prx/pdf/10.1103/PhysRevX.9.031041) [Auto-diff linear algebra](https://arxiv.org/abs/1710.08717) The added test includes a numerical check (via central differences) of the analytical gradient since this is a novel implementation. The tests were run offline 1K times to insure against flakiness. ## Checklist ## - [ ] Changes are complete (i.e. I finished coding on this PR) - [ ] All changes have test coverage: - [ ] Code is well-documented: - [ ] To the best of my knowledge, examples are either not affected by this change, or have been fixed to be compatible with this change ### Changes ### - [ ] QR back part 2 - [ ] Tests ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org
