1
$\begingroup$

Given two square matrices $A$ and $B$ of the same dimensions with entries in $\mathbb{C}$, as well as the sets of all eigenvalues and eigenvectors for both matrices $\{e_A,v_A\}$ and $\{e_B,v_B\}$, is there anything non-trivial we can conclude about the set of eigenvalues and eigenvectors $\{e_{AB},v_{AB}\}$ for the multiplied matrix $A\cdot B$?

$\endgroup$
  • 1
    $\begingroup$ Bhatia's Matrix Analysis lists some useful inequalities to bound both eigenvalues and eigenvector variations. Most of them apply to $A \pm B$, but there are results for $AB$ as well. $\endgroup$ – Omnomnomnom Dec 20 '17 at 20:23
2
$\begingroup$

I don't know how trivial this is to you, but here it is:$$0\in e_{AB}\iff 0\in e_A\vee0\in e_B.$$


Here's another one: if $v\in v_A\cap v_B$, then $v\in v_{AB}$.

$\endgroup$
  • $\begingroup$ This makes sense, thank you! I wonder if there are more properties worth mentioning? $\endgroup$ – Kagaratsch Dec 20 '17 at 18:22
  • 1
    $\begingroup$ @Kagaratsch I've added another one. $\endgroup$ – José Carlos Santos Dec 20 '17 at 18:35
  • $\begingroup$ I am more curious about relations such as writing $v_{AB}$ as a linear combination of $v_A$ or $v_B$ with perhaps a well defined procedure to fix the coefficients... But maybe that is too optimistic. $\endgroup$ – Kagaratsch Dec 20 '17 at 19:29

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.