Ilhan Polat

> This functionality was recently added, in https://github.com/scipy/scipy/pull/22228, so good chance that the problem is on our side. Yes, probably, we missed something. @Mugundanmcw do you have other means of...

I am only wary about the `fwin1` name which tells very little and no 2D but there are too many resolved reviews above so if already discussed nevermind me.

> The `test_lapack` failures are still there unfortunately Yeah, it looks like we are not seeing something obvious that only hits this platform, but not sure what it is yet.

This does not seem like a SciPy issue but the underlying LAPACK vendor might have a bug. I tried many inputs with various size but did not manage to crash...

If there is a way to test the OpenBLAS functionality with a simple C/C++/Fortran program by linking to it would clarify things better on this platform. Regular LAPACK repo is...

Possibly my fault as a bug since I am also hunting for a complex bug too. The real question is how to reproduce these failures since they all pass in...

Could someone please try this patch, the mistake must be related to these branches ```diff --- a/scipy/sparse/linalg/_eigen/arpack/arnaud/src/arnaud_n_double_complex.c +++ b/scipy/sparse/linalg/_eigen/arpack/arnaud/src/arnaud_n_double_complex.c @@ -364,7 +364,8 @@ ARNAUD_zneupd(struct ARNAUD_state_d *V, int rvec, int howmny,...

Sorry I edited after a minute or so but not quick enough I think

Thanks for the confirmation. Then in total this would be the patch both the fix and the changes to remove the warnings. Please feel free to open a PR since...

> real_if_close is not that taxing for a user---and in fact in the vast majority of use cases it's just not needed. Quite the other way around. I think this...