19

I came across one problem in a finite field as follows:

Let $F$ be a finite field. Show that if $a, b\in F$ both are non-squares, then $ab$ is a square.

I wanted to prove it by using the idea of Biquadratic field extension. But there is no biquadratic extension over finite fields. Please, any hints for proving above fact? Thanks.

Aysha A.
  • 680
  • 5
  • 15
  • 1
    What's wrong with using the idea you were going to use in your proof? Even over the rationals, you'd still have to deal with the case that $\mathbb{Q}(\sqrt{a}) = \mathbb{Q}(\sqrt{b})$. –  Sep 23 '16 at 18:04

5 Answers5

32

If the characteristic of $F$ is $2$, then $x\mapsto x^2$ is an automorphism of $F$ and therefore every element is a square. So we can assume the characteristic is an odd prime.

Consider the multiplicative group $F^*$ of nonzero elements; the map $x\mapsto x^2$ is a group endomorphism of $F^*$ with kernel $\{1,-1\}$. Therefore the image $H$ of this map is a subgroup satisfying $|H|=|F^*|/2$; this amounts to saying that $H$, which is the set of all squares in $F^*$, has index $2$. Therefore $F^*/H$ is a two-element group and the statement follows.

egreg
  • 244,946
  • 1
    +1 Because this argument does not need the cyclicity of the multiplicative group! – Jyrki Lahtonen Sep 23 '16 at 17:30
  • @egreg I just want to make sure I understood the end of your argument. Since $F^/H$ is a two-element subgroup, $F^/H={\overline{0},\overline{a} }$ for some $a\in F^$. Then this means that if $x,y \in F^$ are nonsquares then $\overline{xy}=\overline{a}$? What does this imply? – Sarah May 01 '17 at 23:01
  • 2
    @Sarah If $x$ and $y$ are nonsquare, then $xH=yH$, so $(xy)H=(xH)(yH)=(xH)^2=H$, implying $xy\in H$. – egreg May 01 '17 at 23:06
  • @egreg The Frobenious map need not be an automorphism for characteristic 2 field, one needs the field to be finite too. – Babai Feb 09 '20 at 18:40
  • @Babai The question is about finite fields. – egreg Feb 09 '20 at 19:56
  • Oh yes! Sorry, I didn't notice. – Babai Feb 09 '20 at 19:58
11

Hint: I assume, $a,b,\neq 0$ otherwise it is obvious. $F^*$ is cyclic. Suppose it is generated by $x$, you can write $a=x^n, b=x^m$, $n,m$ odd. Then $n+m$ is even.

8

Zero is a square, hence we may assume that both $a$ and $b$ belong to $\mathbb{F}^*$.
$\mathbb{F}^*$ is a cyclic group with order $q-1$: if the characteristic is $2$, every element of $\mathbb{F}$ is a square and there is nothing to prove. Otherwise, we may assume that $\mathbb{F}^*$ is generated by some element $g$, and in such a case the quadratic residues in $\mathbb{F}^*$ are the elements of the form $g^{\text{even}}$ and the non-quadratic residues are the elements of the form $g^{\text{odd}}$. Since $\text{odd}+\text{odd}=\text{even}$, the product of two non-quadratic residues is a quadratic residue, i.e. the Legendre symbol is multiplicative.

Jack D'Aurizio
  • 361,689
  • Here is my trial: Let $F^$ be a multiplicative group of nonzero elements in $F.$ Then $F^$ is a cyclic group. Let $c$ be a generator of $F^.$ Thus, $={c, c^2, ..., c^n=e},$ where $e$ is an identity element of $F^.$ Let $a$ and $b$ be both non-squares elements in $F^*,$ then $a$ and $b$ are odd powers of $c.$ Assume $a=c^{2m+1}$ and $b=c^{2n+1}.$ Now $a.b=c^{2m+1}.c^{2n+1}=c^{2(m+n+1)}=\big[c^{(m+n+1)}\big]^2.$ Hence, $ab$ is a square. – Aysha A. Oct 07 '16 at 12:06
  • 1
    @AyshaA.: that is fine. – Jack D'Aurizio Oct 07 '16 at 12:07
6

Egreg's argument is probably the simplest, followed by using cyclicity of the multiplicative group as outlined by Tsemo and Jack. The following approach might be closer to your first idea - utilizing the fact that a finite field has no biquadratic extensions.

So my answer needs the fact that a finite field $F$ has a unique quadratic extension (up to an $F$-isomorphism).


We are given two elements $a,b\in F$ that don't have a square root in $F$. Therefore both extension fields $F(\sqrt a)$ and $F(\sqrt b)$ are quadratic, and hence isomorphic over $F$. This means that we can find a square root of $b$ inside $F(\sqrt a)$. More precisely, it implies that there exist elements $c_1,c_2\in F$ such that $$ (c_1+c_2\sqrt a)^2=b.\qquad(*) $$ Expanding the left hand side gives $$ (c_1^2+ac_2^2)+2c_1c_2\sqrt a=b. $$ Because $1,\sqrt a$ is a basis for the extension, and the right hand side is an element of $F$ we can conclude that we must have $2c_1c_2=0$. There are three ways how this could happen:

  • If $2=0$, then we are in characteristic two, squaring is injective (see Jack's answer, or infer from Egreg's answer that squaring has trivial kernel), and therefore all the elements of $F$ are squares.
  • If $c_2=0$, then $(*)$ is saying that $c_1^2=b$ contradicting the assumption that $b$ has no square root in $F$.
  • That leaves the possibility $c_1=0$. So there exists an element $c_2\in F$ such that $ac_2^2=b$. This implies that $$(ac_2)^2=a(ac_2^2)=ab$$ and therefore in this case $ab$ is the square of an element of $F$.
Jyrki Lahtonen
  • 140,891
3

For my abstract algebra class (which was rings-first to an extent that I didn't ever get to defining a group...), I have written up a proof that uses neither groups nor field extensions. It is spread across two assignments:

  • Exercise 5 (d) on UMN Spring 2019 Math 4281 homework set #6 shows that if $\mathbb{F}$ is a finite field with $2 \cdot 1_{\mathbb{F}} \neq 0_{\mathbb{F}}$, then $\mathbb{F}$ contains exactly $\dfrac{1}{2}\left(\left|\mathbb{F}\right|+1\right)$ many squares. This is proven by a double counting argument: Every nonzero square is a square of exactly two different elements of $\mathbb{F}$ (the word "different" here relies on $2 \cdot 1_{\mathbb{F}} \neq 0_{\mathbb{F}}$), and the square $0$ is a square of only one element of $\mathbb{F}$.

  • Exercise 4 (c) on UMN Spring 2019 Math 4281 midterm #3 shows that if $\mathbb{F}$ is a finite field with $2 \cdot 1_{\mathbb{F}} \neq 0_{\mathbb{F}}$, then any product of two nonsquares in $\mathbb{F}$ is a nonsquare. This is proven in a somewhat tricky way, by first arguing that a product of squares is a square and that a product of a nonzero square with a nonsquare is a nonsquare, and finally making conclusions by counting (using the previously proven fact that $\mathbb{F}$ contains exactly $\dfrac{1}{2}\left(\left|\mathbb{F}\right|+1\right)$ many squares).

  • Finally, Theorem 4.1 (c) in Section 4.4 of UMN Spring 2019 Math 4281 midterm #3 proves that in any finite field $\mathbb{F}$, the product of two nonsquares is a square. Indeed, if $2 \cdot 1_{\mathbb{F}} \neq 0_{\mathbb{F}}$, then this was already proven; but if $2 \cdot 1_{\mathbb{F}} = 0_{\mathbb{F}}$, then there are no nonsquares in $\mathbb{F}$ to begin with, and the claim is moot.

All of the proofs here use just basic properties of fields and the pigeonhole principle.