Add matrix_rank by vishwakftw · Pull Request #10338 · pytorch/pytorch

vishwakftw · 2018-08-08T01:39:19Z

Similar functionality as NumPy
Added doc string
Added tests

- Similar functionality as NumPy - Added doc string - Added tests

aten/src/ATen/native/LinearAlgebra.cpp

  return V.mm(S_pseudoinv.diag().mm(U.t()));
 }

+static double _get_epsilon(const ScalarType& sc_type) {


aten/src/ATen/native/LinearAlgebra.cpp

+           "of floating types");
+
+  Tensor S = _matrix_rank_helper(self, symmetric);
+  return (S > tol).toType(kLong).sum().toCLong();


vishwakftw · 2018-08-09T04:51:58Z

@ssnl Is this good to go?

aten/src/ATen/native/LinearAlgebra.cpp

+           "of floating types");
+
+  Tensor S = _matrix_rank_helper(self, symmetric);
+  return (S > tol).toType(kLong).sum();


aten/src/ATen/native/LinearAlgebra.cpp

  return V.mm(S_pseudoinv.diag().mm(U.t()));
 }

+static double _get_epsilon(const ScalarType& sc_type) {


aten/src/ATen/native/LinearAlgebra.cpp

+           "of floating types");
+
+  Tensor S = _matrix_rank_helper(self, symmetric);
+  double tol = S.max().toCDouble() * std::max<double>(self.size(0), self.size(1)) *


…nto matrix-rank

facebook-github-bot

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

aten/src/ATen/native/LinearAlgebra.cpp

  return V.mm(S_pseudoinv.diag().mm(U.t()));
 }

+Tensor _matrix_rank_helper(const Tensor& self, bool symmetric) {


aten/src/ATen/native/LinearAlgebra.cpp

+           "of floating types");
+
+  Tensor S = _matrix_rank_helper(self, symmetric);
+  Tensor tol = S.max() * _get_epsilon(self.type());


torch/_torch_docs.py

+
+:attr:`tol` is the threshold below which the singular values (or the eigenvalues
+when :attr:`symmetric` is ``True``) are considered to be 0. If :attr:`tol` is not
+specified, :attr:`tol` is set to `S.max() * max(S.size()) * eps` where `S` is the


torch/_torch_docs.py

+
+    >>> a = torch.eye(10)
+    >>> torch.matrix_rank(a)
+    10


aten/src/ATen/native/LinearAlgebraUtils.h

+static inline Tensor _get_epsilon(const Type& type) {
+  switch (type.scalarType()) {
+    case at::ScalarType::Half:
+      return type.tensor({}).fill_(std::numeric_limits<at::Half>::epsilon());


goldsborough

Can be a future fix.

aten/src/ATen/native/LinearAlgebra.cpp

+
+  Tensor S = _matrix_rank_helper(self, symmetric);
+  Tensor tol = _get_epsilon(self.type()) * std::max(self.size(0), self.size(1));
+  tol.mul_(S.max());


ssnl

Thanks!

torch/_torch_docs.py

+:attr:`tol` is the threshold below which the singular values (or the eigenvalues
+when :attr:`symmetric` is ``True``) are considered to be 0. If :attr:`tol` is not
+specified, :attr:`tol` is set to ``S.max() * max(S.size()) * eps`` where `S` is the
+singular values (or the eigenvalues when :attr:`symmetric` is ``True``), and `eps`


facebook-github-bot

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

vishwakftw · 2018-08-12T01:03:03Z

@ssnl is this good to go?

vishwakftw · 2018-08-15T00:07:04Z

@ssnl ping :)

ssnl · 2018-08-15T00:08:08Z

I tried to get another person to look before merging but wasn't able to. Will get this going tomorrow.

facebook-github-bot

SsnL is landing this pull request. If you are a Facebook employee, you can view this diff on Phabricator.

test/test_torch.py

        self.assertEqual(X, Xhat, 1e-8, 'USV\' wrong')

+    @staticmethod
+    def _test_matrix_rank(self, conv_fn):


…nto matrix-rank

facebook-github-bot

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

vishwakftw · 2018-08-17T16:40:27Z

@ssnl is this good to go?

ssnl · 2018-08-17T17:27:56Z

It’s failing internal test. I’ve no idea why yet, but I’ll look this afternoon.

…

On Fri, Aug 17, 2018 at 12:40 Vishwak Srinivasan ***@***.***> wrote: @ssnl <https://github.com/SsnL> is this good to go? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#10338 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AFaWZb8d-qn1fW7RjAUpfNUKWLmz97vsks5uRvIHgaJpZM4VzIvZ> .

ssnl · 2018-08-19T01:16:29Z

@vishwakftw There is some really weird things going on. I tried but can't quite repro or figure out why. I will talk with people knowing more about those tests on Monday. Sorry about it.

vishwakftw · 2018-08-19T01:29:10Z

Oh no worries. Please take your time. After all, it’s a weekend :)

ssnl · 2018-08-21T20:08:49Z

@vishwakftw the 3rd arg of np.linalg.matrix_rank is added in NumPy 0.14... Can you add a version check on the matrix_rank(aaT.cpu().numpy(), 0.01, True) test case? This is causing the internal test failure. Sorry about it.

facebook-github-bot

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

Summary: - Similar functionality as NumPy - Added doc string - Added tests Differential Revision: D9240850 Pulled By: SsnL fbshipit-source-id: 1d04cfadb076e99e03bdf699bc41b8fac06831bf

Add matrix_rank

b467995

- Similar functionality as NumPy - Added doc string - Added tests

vishwakftw requested review from apaszke, colesbury, ezyang, gchanan, soumith, ssnl, zdevito and zou3519 as code owners August 8, 2018 01:39

vishwakftw commented Aug 8, 2018

View reviewed changes

ssnl reviewed Aug 8, 2018

View reviewed changes

vishwakftw added 2 commits August 9, 2018 08:08

Modify matrix_rank to return scalar instead of int

5989f8a

Update return type of matrix_rank in docs

e1f1012

ssnl reviewed Aug 9, 2018

View reviewed changes

vishwakftw added 2 commits August 9, 2018 13:26

CR

15753e0

Merge branch 'matrix-rank' of https://github.com/vishwakftw/pytorch i…

2374e88

…nto matrix-rank

facebook-github-bot reviewed Aug 9, 2018

View reviewed changes

ssnl reviewed Aug 9, 2018

View reviewed changes

torch/_torch_docs.py Outdated

>>> a = torch.eye(10)

>>> torch.matrix_rank(a)

10

This comment was marked as off-topic.

Sign in to view

CR by SsnL

08466d5

goldsborough suggested changes Aug 9, 2018

View reviewed changes

goldsborough approved these changes Aug 9, 2018

View reviewed changes

ssnl reviewed Aug 9, 2018

View reviewed changes

aten/src/ATen/native/LinearAlgebra.cpp Outdated

Tensor S = _matrix_rank_helper(self, symmetric);

Tensor tol = _get_epsilon(self.type()) * std::max(self.size(0), self.size(1));

tol.mul_(S.max());

This comment was marked as off-topic.

Sign in to view

More CR

fbeecf9

ssnl approved these changes Aug 10, 2018

View reviewed changes

Update _torch_docs.py

a1fe4fc

facebook-github-bot reviewed Aug 10, 2018

View reviewed changes

soumith approved these changes Aug 15, 2018

View reviewed changes

facebook-github-bot reviewed Aug 15, 2018

View reviewed changes

ssnl requested changes Aug 15, 2018

View reviewed changes

test/test_torch.py

self.assertEqual(X, Xhat, 1e-8, 'USV\' wrong')

@staticmethod

def _test_matrix_rank(self, conv_fn):

This comment was marked as off-topic.

Sign in to view

This comment was marked as off-topic.

Sign in to view

vishwakftw added 2 commits August 16, 2018 08:30

Fix CUDA tests

79f815e

Merge branch 'matrix-rank' of https://github.com/vishwakftw/pytorch i…

b4b3324

…nto matrix-rank

ssnl approved these changes Aug 16, 2018

View reviewed changes

facebook-github-bot reviewed Aug 16, 2018

View reviewed changes

check numpy version for test

c9052b5

facebook-github-bot reviewed Aug 22, 2018

View reviewed changes

vishwakftw closed this Aug 23, 2018

vishwakftw deleted the matrix-rank branch August 23, 2018 01:59

ezyang added the open source label Jun 24, 2019

Conversation

vishwakftw commented Aug 8, 2018 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

vishwakftw commented Aug 9, 2018

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

facebook-github-bot left a comment

Choose a reason for hiding this comment

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

goldsborough left a comment

Choose a reason for hiding this comment

Uh oh!

This comment was marked as off-topic.

Uh oh!

ssnl left a comment

Choose a reason for hiding this comment

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

facebook-github-bot left a comment

Choose a reason for hiding this comment

Uh oh!

vishwakftw commented Aug 12, 2018

Uh oh!

vishwakftw commented Aug 15, 2018

Uh oh!

ssnl commented Aug 15, 2018

Uh oh!

facebook-github-bot left a comment

Choose a reason for hiding this comment

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

vishwakftw commented Aug 8, 2018 •

edited

Loading