3
$\begingroup$

Let $A$ be a $n\times n$ matrix with entries on its diagonal are positive and other entries are negative with sum of entries in every column is 1. Prove that $\det(A) > 1.$

I got no idea to begin with. Any suggestion or hint?

$\endgroup$
1
$\begingroup$

By your assumptions, $A^T$ is a matrix with positive entries on the diagonal and negative off diagonal entries such that each row sums to 1. Let $B$ denote any matrix satisfying these conditions, we'll prove that $\det(B)>1$. First, notice that $B$ has an eigenvector $(1,\dots,1)$ with eigenvalue 1. Now suppose $\textbf{x}=(x_1,\dots,x_n)$ is an eigenvector with complex entries, not all the same, and eigenvalue $\lambda$, we'll show that $|\lambda|>1$. Suppose $x_i$ has maximal modulus among the entries of $\textbf{x}$ and suppose WLOG that $x_i>0$ (otherwise just multiply by the appropriate phase). Now we know that $$|\lambda x_i|=|\sum_{j\neq i}b_{i,j}x_j+b_{i,i}x_i| \geq |b_{i,i}x_i|-\sum_{j\neq i}|b_{i,j}x_j| = b_{i,i}x_i+\sum_{j\neq i}b_{i,j}|x_j| \geq \sum_{j=1}^n b_{i,j}x_i=x_i $$ But one of the above inequalities must be a strong inequality (the first one if $x_j$ all have the same modulus and the second if not). This implies that $|\lambda|>1$, as desired.

$\endgroup$
  • $\begingroup$ Spent some time to catch your first ineq, but still struggling. Would you mind to elaborate on it? $\endgroup$ – Hanno Apr 14 at 14:32
  • $\begingroup$ @Hanno No problem, I've added an intermediate step. Sorry if it wasn't clear. $\endgroup$ – Amichai Lampert Apr 14 at 14:53
  • $\begingroup$ Now I got it & you got it too, +1 $\endgroup$ – Hanno Apr 15 at 13:06
  • $\begingroup$ @Hanno haha thanks $\endgroup$ – Amichai Lampert Apr 15 at 13:38

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.