Ilhan Polat
Ilhan Polat
I guess, we should have just switched to boost directly instead of translating them.
It's due to how Fortran77 `SAVE` attribute works which marks the seed in the current arpack. Using C-lingo, the seed is marked as `static` inside the function. ```C void somePACK(this*...
> Hrm, but I don't think we recently changed the way we import SciPy. In fact, this test merely imports SciPy, then runs a reproducibility test; i.e., repeat some operation...
> Maybe I am wrong, I don't think you are wrong at all. This is probably something else. > I also think we should at least be able to achieve...
Yes, SLSQP is done (sans a few test failures) and I started ARPACK hunt, I'm unfortunately not very good at gdb usage so I'm slow making progress. I now have...
I think it is effort better spent to find what went wrong with the existing code than waiting my progress. I'm a bit overwhelmed from ping-ponging undebuggable F77 and stupid...
Apologies for the big delay, the work is limiting my FOSS activities a lot lately. I am now quite confident that I managed to yank out the randomization parts out...
Is this a wheel install or you are building it yourself in the CI?
Do we still need to encode into two letters a lot of meaning? Why not just `(s,d,c,z)skewsymm` ? Or even plain `(s,d,c,z)skew_hermitian` for that matter?
That is true but `kymm`/`kemm` are also not decided by users. Hence there is always some decision priority by the code authors regardless. Moreover users typically don't have any preference...