Let $A \in \mathcal{M}_3(\mathbb{C})$ such that $\mathrm{tr}(A^2) = \mathrm{tr}(A^3) \in \mathbb{Q},$ where $\mathrm{tr}(A)$ is the trace of $A.$ It is possible to prove that $\mathrm{tr}(A^4) \in \mathbb{Q}?$


closed as off-topic by Travis, Especially Lime, John B, Mars Plastic, user21820 Sep 15 at 14:23

This question appears to be off-topic. The users who voted to close gave this specific reason:

  • "This question is missing context or other details: Please provide additional context, which ideally explains why the question is relevant to you and our community. Some forms of context include: background and motivation, relevant definitions, source, possible strategies, your current progress, why the question is interesting or important, etc." – Travis, Especially Lime, John B, Mars Plastic, user21820
If this question can be reworded to fit the rules in the help center, please edit the question.

  • 4
    $\begingroup$ Let $x,y,z$ be the eigenvalues of $A$, may be repeated. Can you write down an expression for $tr(A^n)$ in terms of $x,y,z$? $\endgroup$ – астон вілла олоф мэллбэрг Sep 11 at 4:33
  • 3
    $\begingroup$ $tr(A^{n})= x^n+y^n+z^n$ $\endgroup$ – gaurav saini Sep 11 at 5:00

We fix some $t\in \Bbb C$. Let $a,b,c$ be the complex roots of the equation $$ X^3-6tX^2+18t^2 X -36 t^3\ . $$ Let $e_n$ be the elementary symmetric polynomial of degree $n$ in $a,b,c$.

Let $p_n$ be the Newton symmetric polynomial of degree $n$ in $a,b,c$.

By Vieta, $$\left\{ \begin{aligned} e_1 &= a+b+c = 6t\ ,\\ e_2 &= ab+bc+ca = 18t^2\ ,\\ e_3 &= abc = 36t^3\ . \end{aligned} \right. $$ Let $A$ be the matrix with diagonal entries $a,b,c$. Then, using the Newton identities, $$ \begin{aligned} \operatorname{Trace}(A^2) &=a^2+b^2+c^2\\ &=p_2\\ &=e_1p_1-2e_2\\ &=6t\cdot 6t-2\cdot 18t^2=0\ , \\ \operatorname{Trace}(A^3) &=a^3+b^3+c^3\\ &=p_3\\ &=e_1p_2-e_2p_1+3e_3\\ &=0-18t^2\cdot 6t+3\cdot 36t^3=0\ , \\ \operatorname{Trace}(A^4) &=a^4+b^4+c^4\\ &=p_4\\ &=e_1p_3-e_2p_2+e_3p_1-4e_4\\ &=0-0+36t^3\cdot 6t-0\\ &=216t^4\ . \end{aligned} $$ It is clear that we can chose $t$, such that the last expression is not rational. This gives and explicit counterexample.

Later edit: The general family of cubics with roots $a,b,c$, so that $p_1=p_1(a,b,c)=e_1=e_1(a,b,c)=s$ and $p_2=p_3=p_2(a,b,c) =p_3(a,b,c)=q$, thus parametrized by $s,q$, is obtained by solving $$ \begin{aligned} e_1&=p_1=s\ ,\\ 2e_2 &=e_1p_1-p_2=s^2-q\ ,\\ 3e_3 &=e_2p_1-e_1p_2+p_3=\frac 12(s^2-q)s-sq+q\ ,\\ \end{aligned} $$ and $p_4$ is obtained algebraically as $p_4=e_1p_3-e_2p_2+e_3p_1-4e_4$. (Here, $e_4=0$.) The question "It is still possible to find a counterexample so that $s$ (trace of $A$) is a natural (or even rational) number?" has a negative answer. If both $s,q$ are rational, than $p_4$ is rational, as an algebraic expression in $s,q$.

  • 1
    $\begingroup$ Thanks! Nice solution! It is still possible to find a counterexample even that $\mathrm{tr}(A) \in \mathbb{N}^{*}?$ $\endgroup$ – 674123173797 - 4 Sep 11 at 10:25
  • 2
    $\begingroup$ @674123173797-4 No - by those same Newton identities, $\mathrm{tr}(A^4)$ is a polynomial in $\mathrm{tr}(A^3)$, $\mathrm{tr}(A^2)$ and $\mathrm{tr}(A)$. $\endgroup$ – Steven Stadnicki Sep 11 at 17:09
  • $\begingroup$ @Steven Stadnicki. Yes, I realized this later, and the same argument show that $\mathrm{tr}(A^k) \in \mathbb{Q}$ for all $k \in \mathbb{N}.$ $\endgroup$ – 674123173797 - 4 Sep 12 at 4:09

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