Timeline for Order of eigenvalue problem using c++ Eigen library
Current License: CC BY-SA 3.0
15 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Feb 11, 2015 at 22:09 | comment | added | Kirill | I see you rewrote the question entirely, please don't do that. It invalidates existing answers and someone reading this page won't be able to understand what's going on. If you want to change the question, just accept a helpful answer, if appropriate, and ask a new question. | |
| Feb 11, 2015 at 20:02 | comment | added | Federico Poloni | Your matrix looks Hermitian (assuming all the variables appearing in it are real). Are you sure that the eigenpairs can be complex? Also, I second what @BillBarth is saying: you have never defined what your "preferred" ordering is. | |
| Feb 11, 2015 at 18:09 | comment | added | Bill Barth | This is helpful, but the question remains, how is it that you know the $j=1$ eigenvalue is supposed to be that one? Is the label arbitrary or did something about the physics determine that? Can you employ that knowledge to sort the eigenvalues returned by your eigensolver? | |
| Feb 11, 2015 at 14:32 | comment | added | DaPhil | I rewrote the whole question and hope this makes it crystal clear what I am asking. I is more complex now but I thought at the beginning that a simplification would make it easier to understand. @FedericoPoloni No. | |
| Feb 11, 2015 at 14:31 | history | edited | DaPhil | CC BY-SA 3.0 |
Rewrote the whole thing to make stuff clear.
|
| Feb 11, 2015 at 12:41 | comment | added | Federico Poloni | Are the eigenvalues guaranteed to be real? | |
| Feb 11, 2015 at 11:28 | answer | added | Kirill | timeline score: 4 | |
| Feb 11, 2015 at 10:37 | comment | added | DaPhil | It is ok if I do not know what A and B is, but I need to make sure that no eigenvalues of A end up with the ones of B. If that is what continous means than yes, I need them to be continous. | |
| Feb 11, 2015 at 10:35 | history | edited | DaPhil | CC BY-SA 3.0 |
Further explanations.
|
| Feb 11, 2015 at 10:13 | comment | added | Kirill | I also don't really understand, but are you asking for each of $A$ and $B$ eigenvalues to be continuous on your grid, despite the assignment of labels being arbitrary? | |
| Feb 11, 2015 at 10:05 | comment | added | DaPhil | In general, I cannot say if A is always smaller or larger than B... | |
| Feb 10, 2015 at 15:29 | comment | added | Bill Barth | Right, but is the label A or B arbitrary? How are the bands distinguished? If you can just always say that band A is the largest eigenvalue and B the smallest, does it matter if they swap places? | |
| Feb 10, 2015 at 14:42 | comment | added | DaPhil | Actually, I want to calculate an electronic band structure, so the quantum numbers distinguish the different bands, here A and B. I'm not sure if this is helpful or even what you meant. | |
| Feb 10, 2015 at 14:34 | comment | added | Bill Barth | What distinguishes A from B in your underlying problem? | |
| Feb 10, 2015 at 14:04 | history | asked | DaPhil | CC BY-SA 3.0 |