1

How would you find eigenvalues/eigenvectors of a $n\times n$ matrix where each diagonal entry is scalar $d$ and all other entries are $1$ ? I am looking for a decomposition but cannot find anything for this.
For example:

$\begin{pmatrix}2&1&1&1\\1&2&1&1\\1&1&2&1\\1&1&1&2\end{pmatrix}$

user957
  • 3,387
  • You should have clarified your original question; finding exact eigenvectors / eigenvalues of structured matrices as opposed to finding approximate eigenvectors / eigenvalues of arbitrary matrices are very different problems. – Qiaochu Yuan Aug 09 '10 at 07:47
  • they are different questions. one requires a general process of finding eigenvectors while this is specific to this matrix. – user957 Aug 09 '10 at 08:02

1 Answers1

7

The matrix is $(d-1)I + J$ where $I$ is the identity matrix and $J$ is the all-ones matrix, so once you have the eigenvectors and eigenvalues of $J$ the eigenvectors of $(d-1)I + J$ are the same and the eigenvalues are each $d-1$ greater. (Convince yourself that this works.)

But $J$ has rank $1$, so it has eigenvalue $0$ with multiplicity $n-1$. The last eigenvalue is $n$, and it's quite easy to write down all the eigenvectors.

Qiaochu Yuan
  • 419,620