-1

Hi

I want to compute eig-decomposition of symmetric matrix A in matlab. assume that A = V.D. V^-1(standard form ).

I checked eig(A) and svd(A) but svd give me A = U.s.V which U and V are not orthonormal (U.V != I).

and also there is a approximately same problem for eig(A)

Is there any way that I can find standard form in Matlab?

I would appreciate your solutions.

Captain
  • 193
  • 1
  • 10
  • If you want `U`, `S`, `V` with `U` and `V` orthonormal (unitary), maybe what you need is the singular value decomposition ([`svd`](https://es.mathworks.com/help/matlab/ref/svd.html)) – Luis Mendo Jan 30 '18 at 15:06
  • @Ander Biguri yes it is true! the documentation doesn't say that, but if we assume the documentation instead your comment doesn't give an effective solution and I don't understand why you think my question is not useful – Captain Jan 30 '18 at 19:13
  • @Captain `eig(A)` worked as expected for me. Perhaps you can give us a small example that gives incorrect results. – beaker Jan 31 '18 at 18:52

1 Answers1

-1

If eig(A) computes invertible V, you can use Vinv=V^(-1);.

Thus you obtain A= V * D * Vinv

I would suggest you to read according literature first, then ask further or again. http://www.netlib.org/lapack/lug/node31.html The according book is okay, a really good paper for a general overview is "Eigenvalue computation in the 20th century".

Jay-Pi
  • 343
  • 3
  • 13