Added support for complex input for Cholesky decomposition #44895

IvanYashchuk · 2020-09-17T18:03:04Z

Cholesky decomposition now works for complex inputs.

Fixes #44637.

…_value Modified random_fullrank_matrix_distinct_singular_value to have random singular values

dr-ci · 2020-09-17T18:13:16Z

💊 CI failures summary and remediations

As of commit 2c02725 (more details on the Dr. CI page):

1/1 failures possibly* introduced in this PR
- 1/1 non-CircleCI failure(s)

1 failure confirmed as flaky and can be ignored:

pytorch_macos_10_13_py3_test

ci.pytorch.org: 1 failed

Failed: pr/caffe2-pytorch-linux-bionic-rocm3.7-py3.6-test

This comment was automatically generated by Dr. CI (expand for details).

Follow this link to opt-out of these comments for your Pull Requests.

Please report bugs/suggestions on the GitHub issue tracker or post in the (internal) Dr. CI Users group.

See how this bot performed.

This comment has been revised 42 times.

codecov · 2020-09-18T05:36:30Z

Codecov Report

Merging #44895 into master will increase coverage by 0.00%.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master   #44895   +/-   ##
=======================================
  Coverage   67.83%   67.83%           
=======================================
  Files         384      384           
  Lines       49962    49962           
=======================================
+ Hits        33892    33893    +1     
+ Misses      16070    16069    -1

Impacted Files	Coverage Δ
torch/testing/_internal/expecttest.py	`78.57% <0.00%> (+1.02%)`	⬆️

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 9e5045e...2c02725. Read the comment docs.

torch/testing/_internal/common_utils.py

IvanYashchuk · 2020-09-18T16:35:21Z

@anjali411, regarding your comments about changes random_fullrank_matrix_distinct_singular_value. Now I think those changes are unnecessary and unrelated to current PR, so I removed them, and use random_symmetric_pd_matrix tests pass well.

test/test_torch.py

complex dtype

facebook-github-bot

@anjali411 has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.

anjali411

We should also update the backward definition for cholesky operation and gradcheck test for complex. @IvanYashchuk would you like to create a follow-up PR for that? you can find the gradient formula for cholesky decomposition in https://arxiv.org/pdf/1701.00392.pdf 4.53 and 4.54

IvanYashchuk · 2020-09-22T16:41:11Z

We should also update the backward definition for cholesky operation and gradcheck test for complex. @IvanYashchuk would you like to create a follow-up PR for that? you can find the gradient formula for cholesky decomposition in https://arxiv.org/pdf/1701.00392.pdf 4.53 and 4.54

Yes, I'll do that.

facebook-github-bot · 2020-09-23T16:15:11Z

@anjali411 merged this pull request in 5b20bf4.

Summary: Updated `cholesky_backward` to work correctly for complex input. Note that the current implementation gives the conjugate of what JAX would return. anjali411 is that correct thing to do? Ref. #44895 Pull Request resolved: #45267 Reviewed By: bwasti Differential Revision: D23975269 Pulled By: anjali411 fbshipit-source-id: 9908b0bb53c411e5ad24027ff570c4f0abd451e6

IvanYashchuk added 5 commits September 17, 2020 15:25

Support complex numbers for cholesky_cuda

9b6a777

Added complex dtype test cases for cholesky

e88f171

Fixed error

6ec17d5

Changed test_cholesky to use random_fullrank_matrix_distinct_singular…

2bd633f

…_value Modified random_fullrank_matrix_distinct_singular_value to have random singular values

Tests for complex Cholesky now pass

47767f5

IvanYashchuk changed the title ~~[WIP] Added support for complex input for Cholesky decomposition~~ Added support for complex input for Cholesky decomposition Sep 17, 2020

IvanYashchuk requested a review from anjali411 September 17, 2020 18:03

pytorchbot added the open source label Sep 17, 2020

IvanYashchuk force-pushed the cholesky-complex-cuda branch from 39fb705 to 2b921c1 Compare September 17, 2020 18:25

flake8

7409222

IvanYashchuk force-pushed the cholesky-complex-cuda branch from 2b921c1 to 7409222 Compare September 17, 2020 18:31

IvanYashchuk added 2 commits September 17, 2020 13:35

flake8 again

844fff7

Merge branch 'master' into cholesky-complex-cuda

45b005c

anjali411 reviewed Sep 18, 2020

View reviewed changes

torch/testing/_internal/common_utils.py Outdated Show resolved Hide resolved

anjali411 reviewed Sep 18, 2020

View reviewed changes

torch/testing/_internal/common_utils.py Outdated Show resolved Hide resolved

IvanYashchuk added 2 commits September 18, 2020 11:17

Reverted the change to the singular values

2dda381

Use random_symmetric_pd_matrix for input generation

39e5044

IvanYashchuk force-pushed the cholesky-complex-cuda branch from 10eb62a to 39e5044 Compare September 18, 2020 16:33

Fixed a comment

cc18030

IvanYashchuk added 2 commits September 19, 2020 00:13

Increased atol for test_cholesky_batched

76e93d9

Merge branch 'master' into cholesky-complex-cuda

8cd9f53

IvanYashchuk added module: complex Related to complex number support in PyTorch module: linear algebra Issues related to specialized linear algebra operations in PyTorch; includes matrix multiply matmul labels Sep 19, 2020

ngimel reviewed Sep 21, 2020

View reviewed changes

test/test_torch.py Outdated Show resolved Hide resolved

Modified input generation. Now it should be hermitian pos def matrix for

2c02725

complex dtype

anjali411 approved these changes Sep 22, 2020

View reviewed changes

facebook-github-bot reviewed Sep 22, 2020

View reviewed changes

anjali411 reviewed Sep 22, 2020

View reviewed changes

facebook-github-bot closed this in 5b20bf4 Sep 23, 2020

facebook-github-bot added the merged label Sep 23, 2020

IvanYashchuk deleted the cholesky-complex-cuda branch September 23, 2020 16:17

IvanYashchuk mentioned this pull request Sep 24, 2020

Updated cholesky_backward for complex inputs #45267

Closed

mruberry added the Merged label Oct 28, 2020

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Added support for complex input for Cholesky decomposition #44895

Added support for complex input for Cholesky decomposition #44895

Uh oh!

IvanYashchuk commented Sep 17, 2020

Uh oh!

dr-ci bot commented Sep 17, 2020 •

edited

Loading

Uh oh!

codecov bot commented Sep 18, 2020 •

edited

Loading

Uh oh!

Uh oh!

Uh oh!

IvanYashchuk commented Sep 18, 2020

Uh oh!

Uh oh!

facebook-github-bot left a comment

Uh oh!

anjali411 left a comment •

edited

Loading

Uh oh!

IvanYashchuk commented Sep 22, 2020

Uh oh!

facebook-github-bot commented Sep 23, 2020

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

Added support for complex input for Cholesky decomposition #44895

Added support for complex input for Cholesky decomposition #44895

Uh oh!

Conversation

IvanYashchuk commented Sep 17, 2020

Uh oh!

dr-ci bot commented Sep 17, 2020 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

💊 CI failures summary and remediations

ci.pytorch.org: 1 failed

Uh oh!

codecov bot commented Sep 18, 2020 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

Uh oh!

Uh oh!

IvanYashchuk commented Sep 18, 2020

Uh oh!

Uh oh!

facebook-github-bot left a comment

Choose a reason for hiding this comment

Uh oh!

anjali411 left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

IvanYashchuk commented Sep 22, 2020

Uh oh!

facebook-github-bot commented Sep 23, 2020

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

dr-ci bot commented Sep 17, 2020 •

edited

Loading

codecov bot commented Sep 18, 2020 •

edited

Loading

anjali411 left a comment •

edited

Loading