4

I want to show that $\sum_{x\in \mathbb F_{q}}x^i=0$ if $q-1$ does not divide $i$. Can someone give me a hint?

user26857
  • 53,190

1 Answers1

6

Let $F$ be the field with $q$ elements and let $g$ be a generator of the multiplicative group. Note that for any integer $n$ we have $g^n=1$ if and only if $q-1$ divides $n$ because the order of $g$ is $q-1$. Following Daniel Fischer's suggestion, note that $$g^{i}\sum_{k=1}^{q-1}{g^{ik}}=\sum_{k=1}^{q-1}{g^{i(k+1)}}=\sum_{k=2}^{q}{g^{ik}}=\sum_{k=1}^{q-1}{g^{ik}}$$ The equality holds because the group is cyclic and all we've done is permuted the elements in the sum. But $F$ is a field, and $g^i\neq 1$ because $q-1$ does not divide $i$. There is only one element $x$ such that for some element $y\neq 1$ we have $yx=x$, and that is the only element we can't divide by.

Matt Samuel
  • 59,287
  • Thanks alot for helping me. Probably a shame but I still dont understand why we have $g^{i}\sum_{k=1}^{q-1}{g^{ik}}=\sum_{k=1}^{q-1}{g^{ik}}$. Where does this come from? – Masoun Alhachmi Aug 16 '15 at 21:29
  • Edit: OKay, I understand it, clever :) – Masoun Alhachmi Aug 16 '15 at 21:31
  • Well done. But there's no need to use powers of $g$ in the inner sum for the trick to be apparent. – Jyrki Lahtonen Aug 16 '15 at 22:31
  • @JyrkiLahtonen How would you write it? I'm not always known to explain things in the best way (I was a terrible lecturer). – Matt Samuel Aug 16 '15 at 23:37
  • Oh you linked your answer. – Matt Samuel Aug 16 '15 at 23:38
  • Yeah. Sorry about blowing my own trumpet. I'm a bit allergic to duplicate questions in general. And I've been around this place, and I pay special attention to questions about finite fields in particular :-) The upvote was there, of course. After all, there's no doubt whatsoever that you solved this under your own steam! – Jyrki Lahtonen Aug 17 '15 at 07:55
  • Great answer! Let me ask you a question please? When we can consider the sum $\sum\limits_{\alpha \in \mathbb{F}_q} \alpha^i$ where $i\in \mathbb{Z}$, then how do we define $\alpha^i$ if $i\leq 0$ and $\alpha=0$? – RFZ Jan 16 '23 at 18:34
  • @ZFR If $i=0$, you define it as $1$. If $i<0$ that would be division by $0$. That doesn't occur anywhere in the answer. All the powers are nonnegative. You don't need negative powers anyway for nonzero elements because they occur as positive powers. – Matt Samuel Jan 17 '23 at 00:27
  • @MattSamuel, so you mean if $i=0$, then we define $\alpha^i=1$ even if $\alpha=0$, right? – RFZ Jan 17 '23 at 01:11
  • @MattSamuel, but it does not agree with the formula $\sum\limits_{\alpha\in \mathbb{F}q}\alpha^i=-1$ if $q-1\mid i$. However, $q-1\mid 0$. If we follow what you've suggested then we would get that $\sum\limits{\alpha\in \mathbb{F}_q}\alpha^i=0$. – RFZ Jan 17 '23 at 01:14
  • @MattSamuel, i did not get your last comment. Why 0 is not an option? I explained in the above comment that it does not agree with the formula. What is wrong? – RFZ Jan 21 '23 at 00:30
  • Can you explain my last comment please? – RFZ Jan 22 '23 at 20:08