Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Given an $n\times n$ matrix $C= [c_{ij}]$ which is symmetric (i.e. $c_{ij}=c_{ji}\ \forall i,j$) calculate the determinant of the following matrix (assume $c_{ij} \neq 0\ \forall i,j$):

$$\left(\begin{array}{cccccc} \left(\sum_{i=1}^n\sum_{j=1}^n c_{ij}\right)^2 & \left(\sum_{j=1}^nc_{1j}\right)^2 & \left(\sum_{j=1}^nc_{2j}\right)^2 &\dots & \left(\sum_{j=1}^nc_{nj}\right)^2\\\left(\sum_{j=1}^nc_{1j}\right)^2 & c_{11}^2 & c_{12}^2 & \dots &c_{1n}^2\\ \left(\sum_{j=1}^nc_{2j}\right)^2 & c_{21}^2 & c_{22}^2 & \dots &c_{2n}^2\\ \vdots & \vdots &\vdots & \dots &\vdots \\ \left(\sum_{j=1}^nc_{nj}\right)^2 & c_{n1}^2 & c_{n2}^2 & \dots &c_{nn}^2 \end{array}\right)$$

Remark: For instance when $n=2$ the determinant is equal to $2(c_{11}c_{22}-c_{12}^2)^3$ (note $c_{12}=c_{21}$) which is of a nice form, but it seems hard to generalise this to the $(n+1) \times (n+1) $ case.

share|cite|improve this question
Where is the problem from? – Davide Giraudo Sep 11 '12 at 9:41
It appeared in my research on tight frames of orthogonal polynomials. – user31899 Sep 11 '12 at 10:31
up vote 2 down vote accepted

I belive that a approach by block matrix decompositon give a satisfactory answer. Seting

$$ A=\left(\sum_{i=1}^{n}\sum_{i=1}^{n} c_{ij}\right)^2, \quad v^T= \left(\left(\sum_{j=1}^{n}c_{1j}\right)^2, \dots,\left(\sum_{j=1} c_{nj}\right)^2\right) $$ $$ v=\begin{pmatrix} \left(\sum_{j=1}^{n}c_{1j}\right)^2\\ \vdots \\ \left(\sum_{j=1}^{n}c_{nj}\right)^2 \end{pmatrix}, \quad B = \begin{pmatrix} c_{11}^2&\dots&c_{1j}^2&\dots&c_{1n}^2\\ \vdots& &\vdots& &\vdots \\ c_{i1}^2&\dots&c_{ij}^2&\dots&c_{in}^2\\ \vdots& &\vdots& &\vdots \\ c_{n1}^2 &\dots& c_{nj }^2&\dots&c_{nn}^2 \\ \end{pmatrix} $$ We have $(v^TB)=(Bv)^T$ and $v^TBv\in\mathbb{R}$.

Suppose $A$, $D$, $C$, and $B$ are $n\times n$, $n\times m$, $m\times n$-, and $m\times m$ matrices, respectively. Then

$$\det\begin{pmatrix}A& 0\\ C& B\end{pmatrix} = \det\begin{pmatrix}A& D\\ 0& B\end{pmatrix} = \det(A) \det(B) . $$

This can be seen from the Leibniz formula for determinants or by induction on ''n''. When ''A'' is invertible matrix, employing the following identity

$$ \begin{pmatrix}A& D\\ C& B\end{pmatrix} = \begin{pmatrix}A& 0\\ C& I\end{pmatrix} \begin{pmatrix}I& A^{-1} D\\ 0& B - C A^{-1} D\end{pmatrix}$$

leads to

$$\det\begin{pmatrix}A& D\\ C& B\end{pmatrix} = \det(A) \det(B - C A^{-1} D) .$$

When ''B'' is invertible, a similar identity with $\det(B)$ factored out can be derived analogously,These identities were taken from that is,

$$\det\begin{pmatrix}A& D\\ C& B\end{pmatrix} = \det(B) \det(A - D B^{-1} C) .$$

When the blocks are square matrices of the same order further formulas hold. For example, if $C$ and $B$ commute (i.e., $CB = BC$), then the following formula comparable to the determinant of a 2-by-2 matrix holds:Proofs are given in J.R. Silvester, Math. Gazette, 10 (2000), pp. 460-467, available at $$ \det\begin{pmatrix}A& D\\ C& B\end{pmatrix} = \det(AB - DC).$$

Them $\det(C)= \det(B)\cdot( A-v^TB^{-1}v)$, whit $v^TB^{-1}v, A\in\mathbb{R}$ or $$ \det(C)= \det(B)\left[ \left(\sum_{i=1}^{n}\sum_{i=1}^{n} c_{ij}\right)^2 - \left(\left(\sum_{j=1}^{n}c_{1j}\right)^2, \dots,\left(\sum_{j=1} c_{nj}\right)^2\right) B^{-1} \begin{pmatrix} \left(\sum_{j=1}^{n}c_{1j}\right)^2\\ \vdots \\ \left(\sum_{j=1}^{n}c_{nj}\right)^2 \end{pmatrix} \right] $$ whit $ \left(\left(\sum_{j=1}^{n}c_{1j}\right)^2, \dots,\left(\sum_{j=1} c_{nj}\right)^2\right) B^{-1} \begin{pmatrix} \left(\sum_{j=1}^{n}c_{1j}\right)^2\\ \vdots \\ \left(\sum_{j=1}^{n}c_{nj}\right)^2 \end{pmatrix} \in\mathbb{R}$.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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