0
$\begingroup$

I have a concern regarding primitive matrices. We learned in class several theorems to determine whether or not a matrix is primitive. I'll state them bellow for reference:

  1. If an irreducible non-negative matrix $A$ has $h$ eigenvalues, $\lambda_1, \lambda_2, ..., \lambda_h$ of maximum modulus $|{\lambda_1}|=|{\lambda_i}|, i=1,2,...,h$, then A is called primitive if $h=1$ and imprimitive if $h>1$
  2. A non-negative matrix $A$ is primitive iff some power of $A$ is positive. (i.e. $A^p>0$ for some integer $p>1$).
  3. An irreducible Leslie matrix is primitive iff the birth rates satisfy the following relationship (let greatest common divisor = g.c.d): $$g.c.d.\{i|b_i>0\}=1$$


So with these theorems in mind, I see this inconsistency with them, hopefully someone can point out where I'm misunderstanding one of these theorems.
Ex: If we have the following matrix $$L=\begin{bmatrix} 0 & b_2 & 0 & b_4\\ s_1 & 0 & 0 & 0\\ 0 & s_2 &0 &0 \\ 0 & 0 & s_3 & 0 \end{bmatrix}$$, with $b_i, s_i >0$. This is an irreducible Leslie matrix, and depending on what values of $b_2$ and $b_4$, we can make it so that $g.c.d.\{i|b_i>0\}=1$, example if we set $b_2=3$, and $b_4=4$.
Okay, so by theorem 3, the matrix should be primitive. However, there's no positive integer $p$ such that $L^p>0$, which contradicts theorem 2 (I tried several different values for p, going up to 20 and none of them were strictly positive matrices). I really don't understand this. Any help would be greatly appreciated!

  • 3
    gcd of $\{i: b_i > 0\} \subseteq \{2,4\}$ is never 1?2017-02-11
  • 0
    It can be 1, like I showed in my example above, if I set $b2=3$, and $b4=4$, g.c.d. {3,4}=12017-02-11
  • 2
    $\{i : b_i > 0\}$ is a set of column indices, the value of the $b$'s doesn't matter2017-02-11

0 Answers 0