Does there exist a symmetric matrix $A$ such that $2^{\sqrt{n}}\le |\operatorname{Tr}(A^n)|\le2020 \cdot2^{\sqrt{n}}$ for all $n$?

I think no. The trace of $A^n$ equals $\sum\limits_{i=1}^n\lambda_i^n$ where $\lambda_i$ are the eigenvalues of $A$. Now, if the absolute value of trace of $A$ is bounded below by $2$, then I think the trace of $A^n$ will grow infinitely. Am I right? Thanks beforehand.


As you said, for a symmetric matrix $A$ we have $Tr(A^n) = \sum_i \lambda_i^n$. Now consider two cases for $\lambda_\max = \max_i |\lambda_i|$:

  • $\lambda_\max \le 1$. Then $|Tr(A^n)| \le n$, and which is smaller than $2^{\sqrt n}$ for a sufficiently large $n$.
  • $\lambda_\max > 1$. Then for even $n$ we have $Tr(A^n) \ge \lambda_\max^n$, which is greater than $2020 \cdot 2^{\sqrt n}$ for a sufficiently large $n$.

Let $A\in\mathbb{R}^{d\times d}$ and $A$ is symmetric. The eigenvalues $\lambda_1,\ldots,\lambda_d$ of $A$ are real. For all $n\in\mathbb{N}$ $$Tr(A^n)=\sum_{i=1}^n\lambda_i^n$$

If we assume that for all $n\in\mathbb{N}$ $$2^{\sqrt{n}}\leq Tr(A^n)\leq 2020 \times2^{\sqrt{n}}$$

There exists necessarily $i\in\{1,\ldots,d\}$ such that $|\lambda_i|>1$, because otherwise the sequence $Tr(A^n)$ is bounded.

Then for all $n\in\mathbb{N}$ $$\lambda_i^{2n}\leq 2020\times 2^{\sqrt{2n}}$$ So, for all $n\in\mathbb{N}$ $$ 2n\ln(|\lambda_i|)-\sqrt{2n}\ln (2)\leq\ln(2020)$$ which is impossible since $\lim(2n\ln(|\lambda_i|)-\sqrt{2n}\ln (2))=+\infty$.


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.