Ilhan Polat
Ilhan Polat
Unfortunately it is a blocker. But I checked Sandia Labs' Anasazi/Trilinos (Lehoucq is also an author) but it went over my head very quickly. I don't think it is separable...
Hi @yixuan first of all, on behalf of all community, a true congratulations from me for attacking this problem in such a comprehensive fashion. A very commendable work indeed. First,...
> I would prefer to rely on a library that is maintained in the long term, and its author understands the implemented algorithms. This is true for Spectra. As I...
Let me provide some concrete details what we mean by API breakage, take non-symmetric double precision purifier `dneupd` ```Fortran subroutine dneupd (rvec , howmny, select, dr , di, & z...
`bmat=1` signify that there is a B matrix involved. `bmat=0` is the standard problem. That's mostly due to avoid strange portability issues around bool type of C which is one...
Yes that's fair. I was being lazy.
Thanks @szhorvat good points overall. > Would it be possible to annotate the entries of the ARPACK_arnoldi_update_vars struct with whether they are used for input, output or both? Yes, that's...
Apart from those the Array API rng related errors, this is all passing. I should buy a lottery ticket. I thought I'd never see the day. The basic functionality is...
cc @haampie @cournape now that things are stabilized a bit, a gentle ping in case you are interested.
Among the remaining items, #6827 and related https://github.com/opencollab/arpack-ng/issues/51 look like a genuine ARPACK shortcoming instead of a code problem. Not sure how to prevent it yet but I can also...