Tell me more ×
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.
up vote 1 down vote favorite

If $J$ is a singular matrix, then $(J^T J)^{-1}$ is singular too. I'm trying to prove that $J^T J+\lambda I$ is a singular matrix, where $I$ denotes identity matrix. Any suggestions please? Thanks

share|improve this question
2  
Is $\lambda$ intended to be an eigenvalue? – axblount Aug 24 '12 at 16:16
I answered what is written there, but it all seems a little nonsensical. If you decide to change the question a LOT, then consider posting it as a second question. But don't completely delete your original. – rschwieb Aug 24 '12 at 16:20
1  
"if (stuff) is singular then (inverse) is singular too" - isn't "singular" supposed to be synonymous with "non-invertible"? :) – J. M. Aug 24 '12 at 16:30
@J.M. Nevermind! I interpreted your comment correctly now :) I'm with you, synonymous. – rschwieb Aug 24 '12 at 16:39

2 Answers

up vote 7 down vote accepted

Let $x$ be nonzero such that $Jx=0$. Then $J^\mathrm{T} Jx=0$, so $J^\mathrm{T} J$ is singular, and $(J^\mathrm{T} J)^{-1}$ does not even exist.

As for the second sentence in your question, $J$=the zero matrix and its transpose are certainly singular, but $J^\mathrm{T} J+\lambda I$ will certainly not be if $\lambda\neq 0$.

share|improve this answer
What is x? Is a vector? – Mark Aug 24 '12 at 18:02
1  
@Mark Exactly. You know the characterization that $M$ is singular iff $Mx=0$ for some nonzero vector $x$ right? – rschwieb Aug 24 '12 at 18:10
Yes, thanks! :-) – Mark Aug 24 '12 at 18:28
up vote 0 down vote

Let $J^T J +\lambda I = A$. Then, $J^T J = A-\lambda I$.

Since $J^T J$ is not invertible, then $\lambda$ is an eigenvalue of $A$.

If $A$ has an eigenvalue $\lambda = 0$, then your condition that $J^T J + \lambda I$ is singular will be met, because $\det A = \prod_{i=1}^n \lambda_i$.

share|improve this answer
I don't think "since" is what you mean, but I see the idea you want to mention. Basically any vector $x$ such that $J^T Jx=0$ is an eigenvector of $A$ with eigenvalue $\lambda$. – rschwieb Aug 24 '12 at 16:42
I was using the definition of an eigenvalue $\det (A-\lambda I)=0$. If $ J^T J$ is nonsingular, then the determinant is not zero, and so $\lambda$ is not an eigenvalue. But since $J^T J$ is singular, then $\lambda$ is an eigenvalue. – Arkamis Aug 24 '12 at 16:51
I meant that you can't possibly say "since $J^TJ$ is not invertible" because that is not a hypothesis. It is the conclusion, according to the OP's question. – rschwieb Aug 24 '12 at 17:24
I think you're missing what I'm saying. $J$ is singular so naturally $J^T J$ is also singular. Generally speaking, $\lambda$ is an eigenvalue of $B+\lambda I$ if and only if $B$ (square) is singular. Since $J^T J$ is singular, then $\lambda$ is an eigenvalue of $J^T J + \lambda I$. In other words, $\det J = 0$ implies that $\lambda$ is an eigenvalue of $J^TJ + \lambda I$. – Arkamis Aug 24 '12 at 17:31
OK, yeah I did not know where you were beginning, but that's consistent. – rschwieb Aug 24 '12 at 18:08

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.