Eigensolvers for non-symmetric matrices

[maxscheurer]

According to @mfherbst, the current eigensolver implementations (Davidson, Lanczos) assume symmetric matrices, but they can be adapted to solve eigenvalue problems of non-symmetric matrices (required for EOM etc.).

[mfherbst]

How close to needing this are you @maxscheurer?

[maxscheurer]

I don't really need it for anything urgent, moving the gradients forward seems more important right now. (EOM)-CC is more of a fun playground project 😄

[mfherbst]

Agree. Thanks, than I'll review that next (sorry about the delays).

[maxscheurer]

(sorry about the delays).

No worries at all 👍

[obackhouse]

I don't really need it for anything urgent, moving the gradients forward seems more important right now. (EOM)-CC is more of a fun playground project smile

I actually implemented a pretty simply CCSD solver in my adcc fork here, I wanted to implement EOM-CCSD (for fun) but only realised adcc didn't have non-Hermitian solvers after I'd implemented the ground state 😅

[maxscheurer]

Nice! I have autogenerated code for CC2-LR and EOM-CC, too... also just for fun 😬😂

[mfherbst]

sounds like we should throw stuff together and get CC working in adcc ;)

[mfherbst]

I mean the first attempt does not need to be "good".