Problem: Suppose that $p$ is a prime. Suppose that there is $m \in \mathbb{N}$ such that $p=1+3m$. Define: $$f(x)=(x+1)((x+1)^{2m}+(x+1)^{m}+1) \in \mathbb{Z}/p\mathbb{Z}[x]$$ I would like to prove that if $p \neq 7$ and $p \neq 13$ then $x^m-1 \nmid f(x)$ in $\mathbb{Z}/p\mathbb{Z}[x]$.

Attempt: I observed that $f(x-1)(x^m-1)=x^p-x$. Then I observed that $x^p-x$ has all its roots in $\mathbb{Z}/p\mathbb{Z}$ and that its roots are all distinct. I would like to prove that for $m \geq 6$ then $x^m-1$ and $((x-1)^m-1)$ share at least a root. This is sufficient because then we would have that $x^p-x$ has at least one multiple roots (I mean that it is not simple).

  • $\begingroup$ Just for confirmation, by $\Bbb{Z}_p$, do you mean the field with $p$ elements or the $p$-adic integers? $\endgroup$ Aug 29, 2021 at 9:59
  • $\begingroup$ Are you somehow trying to use the fact that for any prime $p$ and $m$ with $p\nmid m$, $\Bbb{Q}_p$ has a primitive $m$-th root of unity iff $m\mid p-1$? $\endgroup$ Aug 29, 2021 at 10:04
  • $\begingroup$ @ShubhrajitBhattacharya I mean the field with $p$ elements. $\endgroup$ Aug 29, 2021 at 12:29
  • $\begingroup$ Ohh, then please use the symbol $\Bbb{Z}/p\Bbb{Z}$ instead of $\Bbb{Z}_p$. The latter is a standard notation for the $p$-adic integers and it might be confusing. $\endgroup$ Aug 29, 2021 at 18:08
  • 1
    $\begingroup$ @JyrkiLahtonen, exactly! $\endgroup$ Aug 30, 2021 at 10:41

2 Answers 2


An element $a\in{\bf Z}/p{\bf Z}$ is a root of $x^m-1$ if and only is it is a non-zero cube. If it is also a root of $(x-1)^m-1$, then $a+1$ is also a non-zero cube. This leads to a ${\bf Z}/p{\bf Z}$-rational point $P$ on the projective genus $1$ curve $x^3=y^3+z^3$ with the property that $x,y,z\not=0$. Since there are nine points with $x,y,z=0$, Hasse's theorem implies that $P$ exists whenever $p+1-2\sqrt{p}>9$. In other words, when $p>16$.

  • $\begingroup$ I appreciate the answer but unfortunately I do not know the “projective genus 1 curve” neither the Hasse’s theorem. I will give you ‘+1’ but I would like to know if there is an answer which uses only elementary algebra’s arguments. $\endgroup$ Aug 31, 2021 at 11:17

This is an attempt, please point out if there is an error.

Not a complete answer. Too long for a comment

Consider the polynomial $f(x)=(x+1)((x+1)^{2m}+(x+1)^m+1)$

We get two useful identities, inspired by the proceedings of the OP

$$f(x)((x+1)^m-1)=(x+1)^p-(x+1)\tag{1}$$$$f(x-1)(x^m-1)=x^p-x \tag{2}$$ Let, if possible, $x^m-1\mid f(x)$. Then write $f(x)=(x^m-1)g(x)$ for some $g(x)\in\Bbb{F}_p[x]$. Then $$(x^m-1)((x+1)^m-1)g(x)=(x+1)^p-(x+1)\tag{3}$$ $$(x^m-1)((x-1)^m-1)g(x-1)=x^p-x\tag{4}$$

Therefore, to reach a contradiction, as the OP predicted, it's enough to prove that at least any two among the three $((x+1)^m-1),(x^m-1),((x-1)^m-1)$ share a common root (because for any indeterminate T, the polynomial $T^p-T$ in $\Bbb{F}_p[T]$ has exactly $p$ distinct roots). Well, the elements of $\Bbb{F}_p$ which are not a root of any of the three mentioned, are elements of order three in the multiplicative group $\Bbb{F}_p^{\times}$ of order three. If all these roots of them were distinct, then we would have at most $3m+1-3m=1$ element of order $3$. Indeed, let $$x_1,x_2,\ldots,x_m$$ be the roots of $x^m-1$. Then $$x_1+1,x_2+1,\ldots,x_m+1\\x_1-1,x_2-1,\ldots,x_m-1$$ are the roots of of the rest of the two polynomials $(x+1)^m-1$

This argument works in two cases

Case 1 when $(x+1)^m-1$ and $x^m-1$ share a root

Case 2 When $x^m-1$ and $(x-1)^m-1$ share a common root

Case 3 What to do when $(x-1)^m-1$ and $(x+1)^m-1$ share a common root?

and $(x-1)^m-1$ and all these $3m$ elements are in $\Bbb{F}_p$. But this is a contradiction to Sylow's theorem which says that there exists a subgroup of order $3$ and hence at least $2$ elements of order $3$.

  • $\begingroup$ Unfortunately I think that this doesn't really work: you have to prove that either $(x+1)^m-1$ and $x^m-1$ share a root or that $x^m-1$ and $(x-1)^m-1$ share a root, you can't mix up $(x+1)^m-1$ and $(x-1)^m-1$, since they belong to different identities (i.e. 3 and 4). $\endgroup$ Aug 29, 2021 at 19:27
  • $\begingroup$ @FedericoClerici, I have mentioned that this is an attempt. So, I forgot to write the two cases. Sorry. $\endgroup$ Aug 29, 2021 at 19:46
  • 1
    $\begingroup$ Well, now it's fine; it's missing the "hard" part of the problem, i.e. proving that $(x+1)^m-1$ and $x^m-1$ always share a common factor for all $p=3m+1>13$. $\endgroup$ Aug 29, 2021 at 20:38

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.