Skip to content

Only compute one eigenvalue instead of all#25

Open
duetosymmetry wants to merge 2 commits into
BlackHolePerturbationToolkit:masterfrom
duetosymmetry:master
Open

Only compute one eigenvalue instead of all#25
duetosymmetry wants to merge 2 commits into
BlackHolePerturbationToolkit:masterfrom
duetosymmetry:master

Conversation

@duetosymmetry
Copy link
Copy Markdown
Member

This addresses #23. Instead of using Eigenvalues[AMatrix] or Eigensystem[AMatrix] (which switch to dense methods for matrices greater than 100x100), we use Eigenvalues[AMatrix, -1, ...] to sparsely compute only the desired eigenvalue. The correct eigenvalue is chosen by using the Arnoldi method, which allows a Shift argument.

barrywardell
barrywardell previously approved these changes May 27, 2020
@duetosymmetry
Copy link
Copy Markdown
Member Author

I'm still worried about the accuracy of switching to the Arnoldi method — the accuracy depends on the number of eigenvalues/vectors computed, and I don't know the optimal number.

@barrywardell
Copy link
Copy Markdown
Member

I guess this might also need to be linked to the size of the matrix, which is also not particularly well chosen currently.

@barrywardell barrywardell dismissed their stale review August 5, 2022 02:22

This doesn't work as well as expected, as explained by duetosymmetry's follow up comment.

@barrywardell barrywardell linked an issue Mar 16, 2023 that may be closed by this pull request
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Ask Mma to only compute one eigenvalue instead of all

2 participants