Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

If I have the two unitary matrices from the SVD of an m x n matrix (U, V*) and I form a set of new matrices by doing $u_iv_i^H$ (forms an m x n matrix). Assuming $r = min(m, n)$ and my set is $X_1, X_2, ..., X_r$, how can I show that this is an orthonormal set? Is there some property of unitary matrices that I've forgotten? Their rows and columns form an orthonormal basis in $C^n$ but what of the product of two orthonormal vectors?

Furthermore, how do I generalize the coordinates of the original matrix using this orthonormal basis of the vector space?

share|improve this question
    
What are $X_k$, the dyads $u_kv_k^*$? If so, what do you mean by orthonormal (for matrices)? –  copper.hat Dec 13 '12 at 19:29
    
By orthonormal matrices, I mean the set is orthogonal and the matrix norm (generalization of the vector norm) is 1. –  PatternMatching Dec 13 '12 at 19:43
1  
I understand the definition of orthonormal. I was asking what inner product you are using for two matrices? One definition is $\langle A, B \rangle = \mathbb{tr} A^*B$, and with this inner product the $X_k$ are orthonormal. –  copper.hat Dec 13 '12 at 20:08
    
Sorry - yes, the inner product would be as you have specified above. –  PatternMatching Dec 13 '12 at 20:17

1 Answer 1

If you let $X_{ij} = u_i v_j^*$, and define the inner product $\langle A, B \rangle = \mathbb{tr} A^*B$, then you have $\langle X_{ij}, X_{ab} \rangle = \delta_{ia} \delta_{jb}$, from which it follows that the $X_{ij}$ form an orthonormal basis (there are $mn$ of them).

The collection of matrices you have above is a subset of this collection, so they are orthonormal (but not a basis).

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.