Ilhan Polat
Ilhan Polat
I am not getting annoyed at typing or whoever uses it. The point is that statically typing is a decades-old discussion and you are not going to win it just...
I still think 20 is a very small size for a sparse algorithm test though. There might be other problems that can show up when the size grows.
@lobpcg I didn't follow but did we figure out the sorting and added the bigger size matrices?
After the rewrite I can comment on this a bit more informed. The returning eigenvalues are rather collected by ARPACK, that is to say, they show up as the algorithm...
Such a strange bug indeed. I'll take a look.
This is turning out to be an architectural problem, not to say that it's on us, but there is some upstream issues here. On my machine, Win10 and Linux Mint...
I don't see anything typed in the Cython code so the resulting compiled code won't make too much of a difference in terms of performance. Do we have any reason...
I think this can all be a single Python file. Then if any, slow parts can be Cythoned out. That should simplify things a bit. `cdef`'d functions are essentially Python/NumPy...
Typing a numpy array input won't make much difference. The body of the functions are untyped (no `cdef'd variables) so Cython has no extra information to work with to make...
I don't think so. It's less of a change than what is being done here. Converting `cdef`s to `def`s, cleaning the signatures and renaming the file will be sufficient instead...