Do different eigenvector algorithms yield different results?

  • Context: Undergrad 
  • Thread starter Thread starter disillusioned
  • Start date Start date
  • Tags Tags
    Eigenvector
Click For Summary

Discussion Overview

The discussion centers around the question of whether different eigenvector algorithms yield different results, particularly comparing the QL algorithm with implicit shifts from Numerical Recipes to MATLAB's LAPACK routines. Participants also explore the uniqueness of eigenvectors and the implications of algorithmic choices on the results obtained.

Discussion Character

  • Debate/contested

Main Points Raised

  • One participant questions if different eigenvector algorithms produce different results, specifically mentioning QL with implicit shifts versus MATLAB's LAPACK routines.
  • Another participant states that different eigenvalues correspond to linearly independent eigenvectors, but acknowledges that two linearly independent eigenvectors can correspond to the same eigenvalue.
  • A third participant asserts that eigenvectors are not unique and that different algorithmic choices can lead to different sets of eigenvectors, providing an example of scalar multiples of an eigenvector being valid.
  • A fourth participant challenges the concept of uniqueness by referencing the identity map, suggesting that any non-zero vector can be considered an eigenvector, thus complicating the notion of uniqueness.

Areas of Agreement / Disagreement

Participants express differing views on the uniqueness of eigenvectors and the impact of algorithmic choices, indicating that the discussion remains unresolved with multiple competing perspectives.

Contextual Notes

Limitations include the lack of specific details on the algorithms mentioned and the potential dependence on definitions of eigenvectors and eigenvalues. The discussion does not resolve the mathematical implications of the claims made.

disillusioned
Messages
1
Reaction score
0
Hi,

Does different eigenvector algorithm give different result?
eg. using QL with implicit shifts frm (Numerical Recipes) vs Matlab's LAPACK routines?

or anyone knows what method Matlab's LAPACK uses & where i can find the source code in c++?

Are eigenvectors unique?

Thanks!
 
Physics news on Phys.org
I don't know what algoritms you mean and I`m not that familiar with Matlab.

Anyway, different eigenvalues always correspond to linearly independent eigenvectors.
But it is possible to have two linearly independent eigenvectors corresponding to the same eigenvalue.

Hope that helps.
 
disillusioned said:
Are eigenvectors unique?
No they're not. Depending on arbitrary choices made while using whatever algorithm you choose you can end up with a different set of eigenvectors than someone else doing the same problem. As a simple example, if [1,2,3] is your eigenvector and [2,4,6]=2[1,2,3] is your friend's, they're both right (assuming one of them is!).
 
look, take the identity map. then anything is an eigenvector (except zero). so what do you mean by unique"
 

Similar threads

  • · Replies 1 ·
Replies
1
Views
2K
  • · Replies 1 ·
Replies
1
Views
3K
  • · Replies 1 ·
Replies
1
Views
4K
  • · Replies 5 ·
Replies
5
Views
2K
  • · Replies 16 ·
Replies
16
Views
5K
  • · Replies 9 ·
Replies
9
Views
2K
  • · Replies 11 ·
Replies
11
Views
4K
  • · Replies 3 ·
Replies
3
Views
2K
Replies
5
Views
2K
  • · Replies 4 ·
Replies
4
Views
2K