Skip to main content
Notice removed Draw attention by CommunityBot
Bounty Ended with no winning answer by CommunityBot
Notice added Draw attention by Ian Morris
Bounty Started worth 100 reputation by Ian Morris
edited tags
Link
Ian Morris
  • 6.2k
  • 2
  • 31
  • 64
Source Link
Ian Morris
  • 6.2k
  • 2
  • 31
  • 64

Matrix semigroups in which a weighted average of eigenvalues is multiplicative

A problem in fractal geometry requires me to consider matrix semigroups with the following curious property. For a $d \times d$ real matrix $A$ let $\lambda_1(A) \geq \lambda_2(A) \geq \cdots \geq \lambda_d(A)$ denote the absolute values of the eigenvalues of $A$. I am interested in semigroups $\mathcal{S} \subset GL(d,\mathbb{R})$ with the property that for some fixed $t \in (0,1)$ depending only on $\mathcal{S}$, $$\lambda_1(AB)\lambda_2(AB)^t = \lambda_1(A)\lambda_2(A)^t\lambda_1(B)\lambda_2(B)^t$$ for all $A,B \in \mathcal{S}$.

Clearly a sufficient condition for the above property is that the absolute eigenvalues $\lambda_1$, $\lambda_2$ are separately multiplicative in $\mathcal{S}$: that is, $\lambda_i(AB)=\lambda_i(A)\lambda_i(B)$ for all $A,B \in\mathcal{S}$ and $i=1,2$. My question is: is this condition also necessary? Is there any other mechanism which can lead to the above property being satisfied?