Timeline for Diagonalize almost symmetric tridiagonal matrix
Current License: CC BY-SA 4.0
11 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Sep 16 at 16:04 | history | edited | N M | CC BY-SA 4.0 |
Fixed indexing typos
|
| Sep 16 at 14:30 | comment | added | Peter A | Thank you for such a beautifully clear answer and for lending me your time and brain power | |
| Sep 16 at 14:26 | vote | accept | Peter A | ||
| Sep 16 at 13:55 | comment | added | N M | The eigenvector computation should be fixed now. Let me know if anything isn't clear. | |
| Sep 16 at 13:53 | history | edited | N M | CC BY-SA 4.0 |
Fixed eigenvector computation
|
| Sep 16 at 13:45 | comment | added | Peter A | On the non-positivity of the eigenvalues, I found this helpful answer that should allow me to make progress mathoverflow.net/questions/131527/… | |
| Sep 16 at 13:43 | comment | added | Peter A | I use python so the SciPy link is very helpful thank you. You are right that I don't need to know the mechanics of the algorithm itself, but just to be able to get the solution fast in code. If you manage to work out the final step, how to convert the eigenvectors for $M'$ into the eigenvectors for $M$ it would be fantastic! I will also try to make progress myself with simple algebra | |
| Sep 16 at 11:38 | comment | added | N M | Actually what I wrote about using Bunch-Nielsen-Sorensen doesn't quite work because the eigenvalues don't change, but I'm pretty sure this can be fixed. I'll think about it and edit my answer later. | |
| Sep 16 at 11:10 | comment | added | N M | This is implemented in LAPACK and is also available in SciPy. If you want to read about the actual algorithm then look at the papers cited in the LAPACK link. | |
| Sep 16 at 5:02 | comment | added | Peter A | Excellent! What algorithm can I use to diagonalise $M'$ in $O(n^2)$? Thank you | |
| Sep 15 at 22:50 | history | answered | N M | CC BY-SA 4.0 |