reduces precision of mdivide_left test for Intel MKL library

[syclik]

Summary:

Reduces the precision of the mdivide_left test. Hopefully this will pass on the Intel MKL library

Intended Effect:

This has no effect on the code base. Just the test, which was failing with the Intel compiler with MKL library, which we don't have to test against.

How to Verify:

Run tests.

Side Effects:

None that we know of.

Documentation:

N/A

Reviewer Suggestions:

@wds15, can you check to make sure this branch passes tests?

[wds15]

Hi!

Sure, will run tests asap.

Thanks.

Von meinem iPad gesendet

Am 25.07.2015 um 18:05 schrieb Daniel Lee notifications@github.com:

Summary:

Reduces the precision of the mdivide_left test. Hopefully this will pass on the Intel MKL library

Intended Effect:

This has no effect on the code base. Just the test, which was failing with the Intel compiler with MKL library, which we don't have to test against.

How to Verify:

Run tests.

Side Effects:

None that we know of.

Documentation:

N/A

Reviewer Suggestions:

@wds15, can you check to make sure this branch passes tests?

You can view, comment on, or merge this pull request online at:

#130

Commit Summary

Fixes #122. reduces precision of mdivide_left test for Intel MKL library
File Changes

M test/unit/math/rev/mat/fun/mdivide_left_test.cpp (32)
Patch Links:

https://github.com/stan-dev/math/pull/130.patch
https://github.com/stan-dev/math/pull/130.diff
—
Reply to this email directly or view it on GitHub.

[wds15]

I can confirm, that the mdivide tests in question do work now with Intel MKL. However, on this branch the choleskey decompose has some trouble:

[----------] 5 tests from AgradMixMatrixCholeskyDecompose
[ RUN ] AgradMixMatrixCholeskyDecompose.exception_mat_fv
[ OK ] AgradMixMatrixCholeskyDecompose.exception_mat_fv (0 ms)
[ RUN ] AgradMixMatrixCholeskyDecompose.exception_mat_ffv
[ OK ] AgradMixMatrixCholeskyDecompose.exception_mat_ffv (0 ms)
[ RUN ] AgradMixMatrixCholeskyDecompose.mat_1st_deriv
unknown file: Failure
C++ exception with description "cholesky_decompose: Cholesky decomposition of m failed" thrown in the test body.
[ FAILED ] AgradMixMatrixCholeskyDecompose.mat_1st_deriv (2 ms)
[ RUN ] AgradMixMatrixCholeskyDecompose.mat_2nd_deriv
unknown file: Failure
C++ exception with description "cholesky_decompose: Cholesky decomposition of m failed" thrown in the test body.
[ FAILED ] AgradMixMatrixCholeskyDecompose.mat_2nd_deriv (0 ms)
[ RUN ] AgradMixMatrixCholeskyDecompose.mat_3rd_deriv
unknown file: Failure
C++ exception with description "cholesky_decompose: Cholesky decomposition of m failed" thrown in the test body.
[ FAILED ] AgradMixMatrixCholeskyDecompose.mat_3rd_deriv (0 ms)
[----------] 5 tests from AgradMixMatrixCholeskyDecompose (2 ms total)

[----------] Global test environment tear-down
[==========] 5 tests from 1 test case ran. (2 ms total)
[ PASSED ] 2 tests.
[ FAILED ] 3 tests, listed below:
[ FAILED ] AgradMixMatrixCholeskyDecompose.mat_1st_deriv
[ FAILED ] AgradMixMatrixCholeskyDecompose.mat_2nd_deriv
[ FAILED ] AgradMixMatrixCholeskyDecompose.mat_3rd_deriv

3 FAILED TESTS

[syclik]

@wds15, what do you think? Should we merge this pull request first, then work on the choleksy test? Or should we separate that into a separate issue?

[wds15]

I suggest to merge as the original issue is fixed. I guess this problem comes due to the recent changes in the choleskey code. Should i run the tests then once more using the develop branch?

Von meinem iPad gesendet

Am 29.07.2015 um 18:33 schrieb Daniel Lee notifications@github.com:

@wds15, what do you think? Should we merge this pull request first, then work on the choleksy test? Or should we separate that into a separate issue?

—
Reply to this email directly or view it on GitHub.

[syclik]

@wds15, I just merged it. Would you mind creating a new issue with the cholesky decompose issue on it?

Yes, rerun the tests if you can.