Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

specfun CI tests for cpp_double_fp_backend #131

Open
ckormanyos opened this issue Jan 9, 2023 · 13 comments
Open

specfun CI tests for cpp_double_fp_backend #131

ckormanyos opened this issue Jan 9, 2023 · 13 comments
Assignees

Comments

@ckormanyos
Copy link
Member

ckormanyos commented Jan 9, 2023

The purpose of this issue is to get the specfun GHA CI tests running for the cpp_double_fp_backend<> template type.

Instantiate at first for double and see how that goes. If more needed, such as __float128, we will decide as we go.

Cc: @sinandredemption and @cosurgi

@ckormanyos ckormanyos self-assigned this Jan 9, 2023
@ckormanyos
Copy link
Member Author

I will take the first try at this.

@ckormanyos
Copy link
Member Author

ckormanyos commented Jan 25, 2023

We have now specfun running (but failing as expected) on develop in CI for cpp_double_fp< 8-byte-double >.
At the moment 11 files fail.
See detailed comments in #138

Cc: @sinandredemption and @cosurgi

@ckormanyos
Copy link
Member Author

Correction: See #138

@ckormanyos
Copy link
Member Author

We keep moving forward, now with only 5 files failing specfun in CI for cpp_double_fp< 8 -byte-double >.

See also #142

Work continues on edge cases which should eventually get all of specfun passinf for this data type.

Cc: @sinandredemption

@ckormanyos
Copy link
Member Author

Hi Fahad (@sinandredemption) in order to provide very clear and easy-to-understand instances of failing test cases, I have prepared the first of two in a series of failing test case studies.

These can be found in #143 and #144

Cc: @cosurgi

@ckormanyos
Copy link
Member Author

ckormanyos commented Jan 29, 2023

We are now down to two and only two meaningful files with failing test cases in specfun.
See the comments in #146

Cc: @sinandredemption and @cosurgi

@ckormanyos
Copy link
Member Author

... down to two and only two meaningful files with failing test cases in specfun.

In particular, see the build log for #146.

@ckormanyos
Copy link
Member Author

See also comments and build logs from #147

@ckormanyos
Copy link
Member Author

ckormanyos commented Apr 19, 2023

I just posted in the wrong thread for these specfun tests.

See #136

Cc: @sinandredemption

@ckormanyos
Copy link
Member Author

See also #150

@ckormanyos
Copy link
Member Author

ckormanyos commented Dec 20, 2024

Still getting closer to passing all of specfun. See detailed comments from #154, repeated here.

Remaining specfun issues include:

  • gamma_inv (several issues ALL around extremely small/subnormal values),
  • one single exception thrown in elliptic_3,
  • several issues (still) in cyl_bessel_k.

Cc: @sinandredemption and @cosurgi

@ckormanyos
Copy link
Member Author

one single exception thrown in elliptic_3

This has been correected in #155. The read of overly large strings (that will be infinite) did not properly regard the result sign of ${\pm}{\infty}$, and they were all positive. Now the sign is correctly retained.

Also the square root of infinity was improperly given in the eval_sqrt function as NaN, but it should also be infinity.

@ckormanyos
Copy link
Member Author

OK with #157 cyl_bessel_k is finished.

The only specfun with remaining test failures is igamma_inv, which will be handled specifically in a separate issue.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

1 participant