If $D$ is a diagonal matrix, with non-negative eigenvalues, prove that there is a matrix $S$ such that $S^2 = D$

  • $\begingroup$ I really don't know how to go about proving this because there are no specific values. Can someone please help? $\endgroup$ – user115241 Dec 12 '13 at 1:04
  • 3
    $\begingroup$ How would you do it if you knew the values of the diagonal of $D$? $\endgroup$ – user7530 Dec 12 '13 at 1:11
  • $\begingroup$ I don't have the values. $\endgroup$ – user115241 Dec 12 '13 at 1:15
  • $\begingroup$ He's asking you how you would do it if you did.... $\endgroup$ – Don Larynx Dec 12 '13 at 1:25

Since $D$ is a diagonal matrix, its eigenvalues are on the diagonal, which you know are non-negative. You can show, using matrix multiplication that for a diagonal matrix, $S$, $S^2$ is $S$ with the values on the diagonal squared. Since all the diagonal values of $D$ are non-negative, then there must be a matrix $S$ such that $S^2 = D$.

  • $\begingroup$ Thank you for your help. What you're saying makes sense, but how would I show that arbitrarily because I am given no values of the matrix. $\endgroup$ – user115241 Dec 12 '13 at 1:14
  • 1
    $\begingroup$ @user115241: If $D=Diag(d_1,\ldots, d_k)$, let $S=Diag(\sqrt{d_1},\ldots, \sqrt{d_k})$. $\endgroup$ – Cheerful Parsnip Dec 12 '13 at 1:16
  • $\begingroup$ You can say that $D_{ii} = \lambda_i$, where $D_{ii}$ is the $i,i$ entry of $D$. Then define $S$ such that $S_{ii} = \sqrt{\lambda_i}$ and $0$ everywhere else. $\endgroup$ – Christopher Liu Dec 12 '13 at 1:17

You seem to be hung up on the fact that "you weren't given specific values". Well, we know that the matrix is a diagonal matrix, so it has to look something like this: $$ D = \pmatrix{ d_{11}&0&\cdots&0\\ 0&d_{22}&\cdots&0\\ \vdots&&\ddots&\vdots\\ 0&0&\cdots&d_{nn} } $$ Now, we don't know what the values from $d_{11}$ to $d_{nn}$ are, but we know that every other entry of the matrix is zero.

Now, what can we say about the eigenvalues of $D$? As Christopher said, the eigenvalues of $D$ are exactly the diagonal entries $d_{11}$ to $d_{nn}$. Why is this the case? Check the definition of an eigenvalue, and check to see why this has to be true.

Now that we know that $d_{11},\dots,d_{nn}$ are the eigenvalues, we know that they have to be non-negative. Since they are non-negative numbers, they have a non-negative square root. Now, look at the matrix $$ S = \pmatrix{ \sqrt {d_{11}}&0&\cdots&0\\ 0&\sqrt{d_{22}}&\cdots&0\\ \vdots&&\ddots&\vdots\\ 0&0&\cdots& \sqrt{d_{nn}} } $$ What happens when you multiply $S$ by $S$, using the rules for scalar multiplication? If you can't figure it out right away, try it for the $2\times 2$ and $3 \times 3$ version, and see if you can find the pattern.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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