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.

Good morning. Assume that we are given a complex vector space. I know that SVD guarantees a factorization of form $M = USV^{*},$ and the Jordan form gives us a factorization of form $M = PJP^{-1}.$ My question is the following: does this mean that for any given operator $M$ on a vector space we can have a matrix from one basis to another basis that is diagonal (via the SVD), and 'almost diagonal'/jordan via Jordan decomposition? My text material isn't quite clear on this, it seems to suggest that via SVD we can indeed get a diagonal matrix for any operator from some specific basis to other (which follow from the SVD decomposition), with the singular values on the diagonal, and a similar thing for Jordan form -- representing any operator as an almost diagonal matrix if we pick the appropriate bases. Thanks for your clarifications.

share|improve this question
    
"...Jordan form gives us a factorization of form $M=PJP^\ast$." - er, no. That particular Jordan form is possible if $M$ were Hermitian. If not, $P$ is certainly not unitary... and thus the correct Jordan decomposition goes like $M=PJP^{-1}$. –  J. M. Nov 12 '11 at 23:03
    
@J.M.: Sorry that's what I meant. I corrected it. –  Matvey Nov 13 '11 at 0:08
add comment

1 Answer

up vote 0 down vote accepted

Let $A = U \Sigma V^*$ be the SVD decomposition of a square matrix $A$. Then $\Sigma$ is not a representation of the linear transformation represented by $A$. Such representing matrices must be similar to $A$, but $\Sigma$ need not be similar. On the other hand, the Jordan decomposition implies that every linear transformation from a vector space to itself, i.e. every endomorphism of a vector space, can be represented by a unique block diagonal matrix, and block diagonal is the simplest form that we can have in general.

share|improve this answer
add comment

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.