Re: [eigen] std c++ linear algebra proposal

[Gael Guennebaud <gael.guennebaud@xxxxxxxxx>]

Hi,

this has been briefly discussed on this ML a few months ago, and I've also sent them some feedback.

It's still not clear to me where this is going though. In particular, the very first drafts were undecided about whether operators should return expressions or concrete objects. From what I heard, several LEWG members are scared by the numerous gotchas of ET, and they are thus rather in favor of returning concrete objects (implying a huge performance penalty).

For a potential backend, there is this other proposal: http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2019/p1673r1.html, which can be summarized as a low-level C++ interface to BLAS/LAPACK. This similar to SLATE, but a bit more c++-ish though.

Cheers,

Gael

On Sun, Nov 10, 2019 at 8:11 PM Rob Conde <rob.conde@xxxxxxxxxxxxxxxx> wrote:

I'm curious if any thought has been given to:

http://open-std.org/JTC1/SC22/WG21/docs/papers/2019/p1385r3.pdf

Is it an alternative to eigen or another potential backend?

Rob Conde