6
$\begingroup$

Let $\epsilon_1,\ldots,\epsilon_n$ be a sequence of signs and $M(t)$ be the tridiagonal matrix whose diagonal entries are $\epsilon_1 t,\ldots, \epsilon_n t$ and off-diagonal entries equal to $1$. Is it true that the number of real zeroes of $P(t)=\det M(t)$ is equal to the absolute value of $\epsilon_1+\dots+\epsilon_n$?

I checked this numerically on various cases, tried to apply the Sturm algorithm but couldn't prove it. Thanks for any help!

$\endgroup$

1 Answer 1

6
$\begingroup$

For $\epsilon_1=\epsilon_2=-1$ and $\epsilon_3=\epsilon_4=\epsilon_5=1$ you get the counterexample $\operatorname{det}M(t)=t(t - 1)^2(t + 1)^2$.

Another example, with simple real roots, is $\epsilon_1=\epsilon_5=-1$ and $\epsilon_2=\epsilon_3=\epsilon_4=1$ with $\operatorname{det}M(t)=(t - 1)t(t + 1)(t^2 + 1)$.

$\endgroup$
3
  • $\begingroup$ Thanks for this perfect answer, do you have an example with $n$ even? I am specially interested in the case when the sequence is palindromic, that is $\epsilon_k=\epsilon_{n+1-k}$. $\endgroup$ Jun 2, 2023 at 21:17
  • $\begingroup$ @JulienMarché Yes, for $\epsilon_1,\ldots,\epsilon_8=-1, -1, 1, 1, 1, 1, -1, -1$ we have $\operatorname{det}M(t)= (t - 1)(t + 1)(t^3 - t - 1)(t^3 - t + 1)$. $\endgroup$ Jun 2, 2023 at 21:33
  • $\begingroup$ Thanks again, this is very helpful. althought this is not a good news for my problem... $\endgroup$ Jun 3, 2023 at 7:47

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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