The approach minimizes arithmetic and communication complexity.
Multiple folks including @typedfemale and Jim Demmel, Inderjit Dhillon, and Ryan Schenider suggested looking at Jacobi methods for fast eigh.
The blocked Jacobi version looks like a great fit on GPUs
Second problem is now out: dense symmetric eigenproblem A=QΛQT. Solution due on July 15!
We've also enabled ncu profiling for your agents on a @verdacloud cloud box sponsored by our good friends at Brev at @NVIDIAAI
No Digg Deeper questions have been answered for this story yet.
https://web.stanford.edu/class/cme335/lecture7.pdf
https://arxiv.org/abs/2506.03466 paper for the screenshot from Jim, Ryan and others
@_arohan_ wow, i'm in good company!
Multiple folks including @typedfemale and Jim Demmel, Inderjit Dhillon, and Ryan Schenider suggested looking at Jacobi methods for fast eigh.
The blocked Jacobi version looks like a great fit on GPUs
https://arxiv.org/abs/2506.03466 paper for the screenshot from Jim, Ryan and others
Multiple folks including @typedfemale and Jim Demmel, Inderjit Dhillon, and Ryan Schenider suggested looking at Jacobi methods for fast eigh.
The blocked Jacobi version looks like a great fit on GPUs
