Prove or disprove there exists a permutation \sigma such that the (i,\sigma(i))th minor and the (i,\sigma(i))th entry of M are both nonzero for each i

161 Views Asked by At

Let $M$ be a square matrix of full rank $n$. I'm wondering if there exists a permutation $\sigma$ such that the $\left(i,\sigma(i)\right)$-th minor of $M$ and the $\left(i,\sigma(i)\right)$-th entry of $M$ are both nonzero for each $i=1,2,\dots,n$.

My observations: For the first row, we can find an index $j$ such that the $\left(1,j\right)$-th minor of $M$ and the $\left(1,j\right)$-th entry of $M$ are both nonzero, which is deduced from the Laplace expansion and $\operatorname{det}(M)\neq0$. Besides, the statement can work for the case of $n=2$. Now I have no idea about how to prove or disprove the whole statement. Any ideas would be appreciated so much. Thank you!

1

There are 1 best solutions below

0
On BEST ANSWER

This claim would be false due to the following ​counterexample:

$$ \begin{bmatrix} 1 & 0 & 1\\ 1 & 1 & 1\\ 0 & 1 & 1 \end{bmatrix}. $$

For the $1$st row, we have to choose the $3$rd column, since the $(1,1)$-th minor and the $(1,2)$-th entry are $0$. However, for the $3$rd row, we still have to choose the $3$rd column since the $(3,1)$-th entry and the $(3,2)$-th minor are $0$.

Thus in the case, there doesn't exist a permutation $\sigma$ such that the $(i,\sigma(i))$-th minor and $(i,\sigma(i))$-th entry are both nonzero for each $i=1,2,3$.