Are there any sequence given by a recurrence relation: $x_{n+t}=P(x_t,\cdots,x_{t+n-1})$, where $P$ is a positive Laurent Polynomial, satisfy:

  • if $x_0=\cdots=x_{n-1}=1$, then the sequence is only integer;

  • but does not exhibit Laurent Phenomenon ?

What if we allow $P$ to be a rational function?

  • 5
    $\begingroup$ Great question! I assume you want $P$ to have positive coefficients, since otherwise $x_n = \dfrac{x_{n-1} + x_{n-2} - x_{n-3}}{x_{n-4}}$ is a counterexample (indeed, the general formula for $x_8$ is not a Laurent polynomial, but if $x_0 = x_1 = x_2 = x_3 = 1$, then all values $x_i$ equal $1$). $\endgroup$ – darij grinberg Feb 22 at 20:07
  • $\begingroup$ @darij grinberg, thanks, I forgot the positivity $\endgroup$ – Wenze 'Sylvester' Zhang Feb 22 at 20:26

Consider $x_{n+3} = \frac{x_n+x_{n+1}}{x_{n+2}}$.

With $x_1=x_2=x_3 =1$ this gives the sequence $1,1,1,2,1,3,1,4,1,5,...$ but it certainly is not Laurent because for instance we have $x_5=\frac{x_2x_3+x_3^2}{x_1+x_2}$.

  • $\begingroup$ Uhm, I wouldn't allow a $2$ in the denominator for a Laurent polynomial -- we're talking integers, right? $\endgroup$ – darij grinberg Feb 22 at 20:35
  • 4
    $\begingroup$ Nice counterexample!! $\endgroup$ – darij grinberg Feb 22 at 20:37

This reminds me of a question I had seen on both MO and MSE. Sequence A276175 in the OEIS is defined by $$a_n = \frac{(a_{n-1} + 1)(a_{n-2}+1)(a_{n-3} + 1)}{a_{n-4}}$$ with $a_0 = a_1 = a_2 = a_3 = 1$. The OEIS page conjectures it to be an integer for all $n$. The MSE question contains a proof the all $a_n$ are integer (though I haven't read the proof). In the comments of the MO question it is observed $a_8$ is not Laurent.

Added in edit: I offer another example which does not exhibit the Laurent phenomenon, but conjecturally is an integer sequence. Consider the sequence defined by $$b_n = \frac{(b_{n-1} + 1)(b_{n-2} + 1)(b_{n-3} + 1)(b_{n-4}+1)}{b_{n-5}}$$ where $b_0 = b_1 = b_2 = b_3 = b_4 = 1$. I computed and after reducing I found the denominator of $b_{10}$ to be $b_0^14(b_1 + 1)b_1^8(b_2 + 1)b_2^4b_3^2b_4$ (not a monomial). I verified this sequence to be integer up to $n=36$, and Kevin O'Bryant later verified up to $n=41$. I asked if the sequence is integer for all $n \geq 0$ in a separate question.

  • 1
    $\begingroup$ Great example! Suggests no general result here... $\endgroup$ – Sam Hopkins Feb 22 at 21:50

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.