Skip to content

Conversation

@cyancirrus
Copy link
Owner

Way more complicated than i first estimated but going to use a design of the following

QR decomp (here householholder) -> Shur -> Givens rotations,
but good to get some numerical compute underneath the fingers

autumn added 30 commits February 19, 2025 18:06
…V in SVD, but have most of the structure up, givens rotation is working as expected as well! a bit of progress, want to get to SVD and then do small refactor, and then move on to the discrete fourier transform
…V in SVD, but have most of the structure up, givens rotation is working as expected as well! a bit of progress, want to get to SVD and then do small refactor, and then move on to the discrete fourier transform
autumn added 10 commits March 5, 2025 18:25
…ther rows and looks useful for other algorithms, just need to relook at algos, diagonalization isn't unique aparently
…!! lets goooooooo!!!, need to rewrite and check and ensure scales and all of that but wowwwwwwwwwwwwwwwwwwwwwww also need to handle complex eigenvalue cases but hey we are doing one step at a time!
… double-diagonalization method, going to implement one from my book and then see if we don't get things as expected, the algorithm given there is like for krysolov subspaces and that one hard fails a simple [1,2,3],[4,5,6],[7,8,9] matrix, so going to implement one a bit more primitie which should isolate if this is numerical instability or if the diagonalization algo needs many more 'hotfixes' like pivoting reortho etc
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.

2 participants