If $\lambda$ is the eigenvalue of matrix $A$,what is the eigenvalue of $A^TA$?I have no clue about it. Can anyone help with that?
Asked
Active
Viewed 143 times
1
-
real matrix...complex matrix...? Not that it matters, but it makes the question more neatly posted. – DonAntonio Apr 19 '14 at 16:51
-
2Do you mean $\lambda$ is one eigenvalue or the only eigenvalue? In general, the eigenvalues of $A^\top A$ have little to do with the eigenvalues of $A$. See singular values. – Ted Shifrin Apr 19 '14 at 16:59
-
@TedShifrin what about $\lambda_{\mbox{max}}A^TA$ and $\lambda_{\mbox{max}}(A)$ – 89085731 Apr 19 '14 at 17:27
2 Answers
2
The good exercise for you is to prove, in general, that $$\lambda_{\text{max}}(A)\le \max_{\|x\|=1} \|Ax\|=\sqrt{\lambda_{\text{max}}(A^\top A)}.$$
Ted Shifrin
- 115,160
-
Is $\frac{(\frac{1}{2}\lambda_{\mbox{min}}(A+A^T))^2}{\lambda_{\mbox(max)}A^TA}<1$ if A is positive definite – 89085731 Apr 19 '14 at 19:26
0
Generally, you won't be able to say much about them.
However, if $A$ is for instance real symmetric, it is diagonalizable with real eigenvalues, meaning there is an orthogonal matrix B (that is with $B^{-1}=B^T$) and a diagonal matrix D such that: $$A = B D B^T$$ $$A^TA = (B D B^T)^T B D B^T = B D^T B^T B D B^T = BD^2B^T$$ In other words, the eigenvalues of $A^TA$ are the squares of the eigenvalues of $A$.
Klaas van Aarsen
- 5,908