Problem: Let $A$=[$a_{ij}$]$_{n x n}$ where $a_{ii}$=$1$, $i=\overline {1,n}$, $a_{ij}=a\not=1, i\not=j$. Find $A^n$, $n\in \mathbb N$.

I need a little help solving this problem. Now, I know how to find $A^n$ if this was matrix 2 by 2, or 3 by 3. I know there is a way to find the eigenvalues and the eigenvectors and form the matrices $P$ (made of eigenvectors) and $D=diag(\lambda_1,...,\lambda_n)$ using $A^n$=$PD^n$$P^{-1}$ (correct me if I am wrong). However, I do not know how to find the eigenvalues and eigen vectors. So, I tried this:

My matrix $A$ would have this form:$$A= \begin{bmatrix} 1 & a & a & \cdots &a \\ a & 1 & a &\cdots &a \\ \vdots & \vdots &\vdots &\ddots&\vdots\\ a & a & a &\cdots &1 \\ \end{bmatrix} $$

So, that is symmetric matrix. If I look at my rows, I see that sum of all elements in each row is $(n-1)a+1$. Could I then say that my eigenvalue is $(n-1)a+1$ ? I was thinking about using $A^n\overrightarrow v=\lambda^n \overrightarrow v$, where $\overrightarrow v$ is my eigenvector, to get the result but I am really not sure how to go from here.

Any help is greatly appreciated.

  • $\begingroup$ Yes you can indeed say that is an eigenvalue, using the fact that for $v=\begin{bmatrix} 1\\1\\\vdots\\1\end{bmatrix}$ $$Av=((n-1)a+1)v$$. Think of other eigenvectors to find them all. But, in my opinion, the best way would be solving the equation $$\det(A-\lambda I)=0$$, as the roots $\lambda$ would be the eigenvalues. In particular, this new matrix looks a lot like yours, only the diagonal entries are different. $\endgroup$
    – Emre
    Jun 2, 2016 at 1:03

1 Answer 1


Continuing from the comment I provided:

$$A-\lambda I=\begin{pmatrix} 1-\lambda&a&\cdots&a\\ a&1-\lambda&\cdots&a\\ \vdots&\vdots&\ddots&\vdots\\ a&a&\cdots&1-\lambda \end{pmatrix} $$

We want this matrix to be singular. We already know that making the diagonals $-(n-1)a$ makes it singular. Trying out for $n=2$, one can immediately see that making all the diagonals equal to $a$ also works. This second possibility corresponds to the eigenvalue $1-a$.

These are the only eigenvalues, as $v_i=(1,0,\ldots,0,\overbrace{-1}^\text{i-th place},0,\ldots,0)$ are for $i=2,\ldots, n$ are the eigenvectors for $\lambda=1-a$.

So, we have $$P=\begin{pmatrix} 1&1&1&\ldots&1\\ 1&-1&0&\ldots&0\\ 1&0&-1&\ldots&0\\ \vdots&\vdots&\vdots&\ddots&\vdots\\ 1&0&0&\ldots&-1 \end{pmatrix}\qquad D=\begin{pmatrix} (n-1)a+1&0&0&\ldots&0\\ 0&1-a&0&\ldots&0\\ 0&0&1-a&\ldots&0\\ \vdots&\vdots&\vdots&\ddots&\vdots\\ 0&0&0&\ldots&1-a \end{pmatrix}$$

After some calculation, one can see that $$P^{-1}=\frac1n\begin{pmatrix} 1&1&1&\ldots&1\\ 1&1-n&1&\ldots&1\\ 1&1&1-n&\ldots&1\\ \vdots&\vdots&\vdots&\ddots&\vdots\\ 1&1&1&\ldots&1-n \end{pmatrix}$$

Let $c=(1-a)^n$ and $b=((n-1)a+1)^n$. Then,

$$A^n=PD^nP^{-1}=\frac1n\begin{pmatrix} b+(n-1)c&b-c&b-c&\ldots&b-c\\ b-c&b+(n-1)c&b-c&\ldots&b-c\\ b-c&b-c&b+(n-1)c&\ldots&b-c\\ \vdots&\vdots&\vdots&\ddots&\vdots\\ b-c&b-c&b-c&\ldots&b+(n-1)c \end{pmatrix}$$

  • $\begingroup$ Thank you very much! I have one more question. After I have $P$ and $D$, do I continue to find $P^{-1}$, and then calculate $P$ $D^n$ $P^{-1}$ or is there some better way to find $A^n$? $\endgroup$
    – Asleen
    Jun 2, 2016 at 17:45
  • 1
    $\begingroup$ My first attempt would be to calculate $A^n$ for $n=1,2,3,4$ and to figure out the pattern. But, calculating $D^N$,$P^{-1}$ and multiplying them is not that hard either. $\endgroup$
    – Emre
    Jun 2, 2016 at 17:50
  • $\begingroup$ I edited the solution, hope this helps. $\endgroup$
    – Emre
    Jun 2, 2016 at 18:27

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.