Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Assuming that characteristic polynomial $\chi_A$ has roots $\lambda_1, \cdots, \lambda_n$ only of odd power, prove that if $P(\lambda_i) = Q(\lambda_i)$ for all $i \in [1,\cdots,n]$ then $P(A) = Q(A)$.

(This is happening in a vector space over an algebraically closed field).

I don't really have much ideas but I think I have to use Jordan normal form or Caley-Hamilton theorem, I am not sure how to use them tho.

Can you please give me a hint on how to start proving this?

EDIT: it looks like my conjecture is false, sorry for the inconvenience.

share|cite|improve this question
up vote 3 down vote accepted

Seems to be false. Try $$A=\pmatrix{2&1&1\cr0&4&2\cr0&-2&0\cr}$$ I believe the only eigenvalue is 2, of multipicity 3. Let $P(x)=x$, $Q(x)=2x-2$. Then $P$ and $Q$ agree at 2, but $P(A)\ne Q(A)$.

share|cite|improve this answer
An easier example would be $A$ a $3 \times 3$ nilpotent matrix and $P=0$, $Q=x$. – Dan Petersen Nov 21 '11 at 7:30

I think what you mean is this. Suppose $A$ is an $n \times n$ matrix whose characteristic polynomial $\chi_A(x)$ has $n$ distinct roots $\lambda_i$ (which are thus eigenvalues of $A$). If $P$ and $Q$ are polynomials with $P(\lambda_i) = Q(\lambda_i)$ for all $i$, then $P(A) = Q(A)$. This is because $P(x) - Q(x)$ is divisible by $\prod_{i=1}^n (x - \lambda_i) = \chi_A(x)$, and Cayley-Hamilton says $\chi_A(A) = 0$.

In the more general case where root $\lambda_i$ has multiplicity $m_i$, what you want in order to conclude $P(A) = Q(A)$ is not just $P(\lambda_i) = Q(\lambda_i)$ but $P^{(j)}(\lambda_i) = Q^{(j)}(\lambda_i)$ for all nonnegative integers $j < m_i$.

share|cite|improve this answer
Except that for that implication you also need characteristic 0. Think about the companion matrix of $X^3$ over a field of characteristic 2, with $P=0$, $Q=X^2$. – Marc van Leeuwen Nov 21 '11 at 10:48
True. OK, the criterion in general is simply that $P(x) - Q(x)$ is divisible (as a polynomial over the field) by $\chi_A(x)$. – Robert Israel Nov 22 '11 at 4:29

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.