$$M=\left[\begin{array}{ccccc} kI & A & \ldots & A & A\\ A & kI & \ldots & A & A\\ \vdots & & & \vdots & \vdots\\ A & A & \ldots & A & kI \end{array}\right]$$ $M$ is a square matrix That is, $M$ has the $kI$ matrix along its main diagonal and matrix $A$ elsewhere.
$A$ is a matrix of dimension $n\times n$ having all elements equal, $kI$ is also an $n\times n$ matrix, the identity matrix multiplied by a constant $k$

How can I find the determinant efficiently of such matrices for small values of $n$? Can it be done by simple transformations??

  • $\begingroup$ @par yes i am sorry for the bad display $\endgroup$ – sashas May 11 '13 at 1:43
  • 2
    $\begingroup$ user77146: Thanks. I deleted my comment by accident but it looks like @Stahl salvaged the LaTeX, so all is well! $\endgroup$ – parsiad May 11 '13 at 1:45

Suppose $M$ is $mn\times mn$. If $A$ is diagonalisable to $D=\operatorname{diag}(\lambda_1,\ldots,\lambda_n)$, then $M$ is similar to $$ M_1=\begin{bmatrix} kI & D & \ldots & D & D\\ D & kI & \ldots & D & D\\ \vdots & & & \vdots & \vdots\\ D & D & \ldots & D & kI \end{bmatrix} $$ and $M_1$ is permutation-similar to $B_1\oplus\cdots\oplus B_n$, where $$ B_i=\begin{bmatrix} k & \lambda_i & \ldots & \lambda_i & \lambda_i\\ \lambda_i & k & \ldots & \lambda_i & \lambda_i\\ \vdots & & & \vdots & \vdots\\ \lambda_i & \lambda_i & \ldots & \lambda_i & k \end{bmatrix}. $$ Hence $$\det(M)=\prod_{i=1}^n\det(B_i)=\prod_{i=1}^n (k-\lambda_i)^{m-1}(k+(m-1)\lambda_i).\tag{1}$$ Since diagonalisable matrices are dense in the matrix space, formula $(1)$ applies for nondiagonalisable $A$ as well. (I assume that you are working over $\mathbb{C}$. For other fields, that $(1)$ works for nondiagonalisable matrices has to be justified by other more general reasons.)


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.