Good evening; can you help me with the problem?

Let an $ n\times n$ matrix A have eigenvalues $\alpha_{1},...\alpha_n$ and eigenvectors ${a}_{1},.. {a}_{n}$ Find eigenvalues and eigenvectors for operator $$L:{Mat}_{n\times m} \to {Mat}_{n\times m}, \; L(X) = AX;$$

I tried to present $$ x = v *u^T ,$$ where $v$ is $n\times n$ and $u^T$ is $n\times m$ , but what to do next, I don't understand.

I looked for examples of similar problems, but I found nothing.

  • 1
    $\begingroup$ Hint: Consider the matrix $X$ whose $i^{\text{th}}$ column is $a_j$ and the remaining entries are all $0$. $\endgroup$ – Michael Burr Feb 19 at 13:29
  • $\begingroup$ @JoséCarlosSantos I don't think he did. You can put any eigenvector in any column. $\endgroup$ – Theo Bendit Feb 19 at 13:59

Let $X$ be an eigenvector for $L$, i.e. a non-zero matrix such that $$AX=\lambda X.$$ Write $X=(x_1,\dots, x_n)$, whith columns $x_1,\dots, x_n$. The above equation can then be read column-wise as $$Ax_i=\lambda x_i.$$ So you see that every column of $X$ has to be an eigenvector of $A$ with corresponding eigenvalue $\lambda$ or the zero vector.

  • $\begingroup$ I believe the question is asking for eigenvectors/values for the operator on the space of matrices. In other words, finding scalars $\lambda$ and non-zero matrices $X$ such that $L(X) = AX = \lambda X$. $\endgroup$ – Theo Bendit Feb 19 at 13:57
  • $\begingroup$ @TheoBendit Thanks for pointing out. I was reading to careless.. $\endgroup$ – James Feb 19 at 14:02

Let $A$ and $X=\begin{bmatrix}x_1&x_2&\dots&x_m\end{bmatrix}$ be as you defined, where $x_i$ is the $i^{\text{th}}$ column of $X$. Then, recall that $$ AX=\begin{bmatrix}Ax_1&Ax_2&\dots&Ax_m\end{bmatrix}. $$

An eigenvector for the operator you defined is a nonzero matrix $X$ so that $AX=\lambda X$ for some $\lambda$.

Let $X_j^i$ be the matrix consisting of all zeros except that vector $a_j$ is in column $i$. In other words, $$ X_j^i=\begin{bmatrix}0&0&\dots&0&a_j&0&\dots&0\end{bmatrix}. $$ Then, \begin{align} AX_j^i&=\begin{bmatrix}A0&A0&\dots&A0&Aa_j&A0&\dots&A0\end{bmatrix}\\ &=\begin{bmatrix}0&0&\dots&0&\lambda_ja_j&0&\dots&0\end{bmatrix}=\lambda_jX_j^i. \end{align}

Therefore, $X_j^i$ is an eigenvector for this map, are these all of the eigenvectors/eigenvalues (how many eigenvalues do you expect and how many have you found)?

  • $\begingroup$ I also have to consider that multiplicity of eigenvalues is possible? $\endgroup$ – GIFT Feb 19 at 20:58
  • $\begingroup$ I'm not sure of your question. Multiple eigenvalues in $A$ is not a problem because you have a full set of eigenvectors. You will certainly have repeated eigenvalues for the given operator, but they have distinct eigenvectors. $\endgroup$ – Michael Burr Feb 19 at 22:11
  • $\begingroup$ I expect n eigenvalues, as I understand i=1....n and $a_j$ are different $\endgroup$ – GIFT Feb 20 at 9:50
  • $\begingroup$ Since $m\times n$ matrices have $mn$ entries, this linear map can be thought of as a map on $\mathbb{R}^{mn}$. This leads to many more eigenvalues, although there are repeats. $\endgroup$ – Michael Burr Feb 20 at 10:37
  • $\begingroup$ Thanks a lot for your help $\endgroup$ – GIFT Feb 21 at 10:50

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.