1

This is a pretty dumb question considering this is the very first question in my exercises (on a section about finite fields). First, the the group of units $\mathbb F_{11}^*$ is $\mathbb{Z}/10\mathbb{Z} = \{ 0,1, \ldots \}$ (it's not how it's defined but it's isomorphic ok). I know that I need to find an element $a$ of maximum order $n$ in $\mathbb{Z}/10\mathbb{Z}$ because then $\mathbb F_{11}^* = \mathbb{Z}/10\mathbb{Z} = \langle a \rangle$, so $a$ is a primitive root.

But I have no idea how to find such an element of maximum order. I can only think of brute force, which doesn't seem like a very efficient way to solve this. Can someone help me?

iwjueph94rgytbhr
  • 932
  • 5
  • 15
  • For any element you choose, you only need to rule out orders $2$ and $5$. – Robert Shore Dec 10 '23 at 03:45
  • 1
    In general, if you consider the group $\mathbb{Z}/m\mathbb{Z}$, an element $a$ is a primitive element if and only if $(a,m)=1$. –  Dec 10 '23 at 04:24
  • You should try and find the primitive on the $\Bbb{F}{11}^*$ side. Finding a primitive element is the way to produce that isomorphism $\Bbb{F}{11}^*\simeq\Bbb{Z}{10}$ as an isomorphism maps a generator to a generator. Trivially $\Bbb{Z}{10}=\langle 1\rangle$, but that won't give you a primitive element for $\Bbb{F}_{11}$. – Jyrki Lahtonen Dec 10 '23 at 04:25
  • 2
    The result in a comment by ShyamalSayak implies that there should be exactly four primitive elements. You can immediately rule out $1$ and $-1$. Actually you can also rule out all the squares, but I'm not sure you have the tools to see that, yet. Anyway, the odds are on your side. Just try some other $a$ :-) – Jyrki Lahtonen Dec 10 '23 at 04:30
  • 2
    Mind you, $\Bbb{F}{11}^$ most emphatically IS NOT EQUAL TO* $\Bbb{Z}{10}$. There multiplicative group $\Bbb{F}{11}^$ is isomorphic to* the additive group $\Bbb{Z}{10}$, but in order to write down that isomorphism you first need to find a primitive element. – Jyrki Lahtonen Dec 10 '23 at 04:33
  • For example $5$ is a primitive element of $\Bbb{F}_{23}^*$ because $5^0\equiv1$, $5^1\equiv5$, $5^2\equiv2$, $5^{3}\equiv10$, $\ldots$, $5^{21}\equiv14$, $5^{22}\equiv1$. All the congruences modulo $23$, and the point is that all the possible remainders $1,2,\ldots,22$ modulo $23$ appeared on the right hand side of the congruences. Albeit in the strange order $1,5,2,10,\ldots,14$. I picked $23$ for this example because in that case $5$ is the smallest primitive root. When the modulus is $11$ you get away with a smaller $a$. Hop to it! – Jyrki Lahtonen Dec 10 '23 at 04:39
  • @FranAguayo You are burying a heap of details under the phrase how to multiply in polar coordinates. If this were remotely feasible modulo all primes, you would completely solve the discrete logarithm problem in $\Bbb{F}_p^*$ for all primes $p$. In other words, care to elaborate?! – Jyrki Lahtonen Dec 10 '23 at 06:02
  • A very relevant answer. See also this. I used the same trick here. – Jyrki Lahtonen Dec 10 '23 at 06:09
  • @JyrkiLahtonen Can you please check my answer, it had a mistake previously which I have fixed. I just want someone to reevaluate the thought process I present for if there's any mistake... – Nothing special Dec 10 '23 at 08:40

4 Answers4

1

You can try the process of elimination.

$$\mathbb F_{11}^* = \{1, 2, \ldots, 10\}$$

Eliminate the identity and the obvious perfect powers (perfect squares and the perfect 5-th powers only because the order of this group is $2\times 5$) Also eliminate $10$ because it has an order of $2$. $10^2\equiv (-1)^2\equiv 1\pmod{11}\tag*{}$

The potential candidates for being generator are $2,3,5,6,7, 8\tag*{}$

$\mathbb F_{11}^*$ admits $\varphi(10)=4$ generators so in our last stage of elimination, exactly two of these numbers will be eliminated.

Claim 1: Either $5$ or $6$ is not a generator.

Hint: $6\equiv -5\pmod{11}$
If $x$ is a generator i.e., $\DeclareMathOperator{\ord}{ord}\ord(x)=10$ then $\ord(x^5)=2\implies x^5\equiv -1\implies (-x)^5\equiv 1\pmod{11}\tag*{}$ $\because$ $-1\pmod{11}$ is the only element of order $2$.

Claim 2: Likewise, either $8$ or $3$ is not a generator.

This means that $2$ must have been a generator (because $2$ is not among the candidates which will be potentially eliminated) and hence, $8=2^3$ is also a generator. Mind that this is true because $\text{gcd}(3,10)=1$.

Thus, $3$ is not a generator and $\ord(3)\neq 2, 10$ so the only divisor of $10$ which remains is $5$, so $\ord(3)=5$.

Now, $6^5=2^5\times \underbrace{3^5}_{\equiv 1} \equiv -1\pmod{11} \tag*{}$ Thus, $\ord(6)=10$ i.e., $6$ is also a generator and hence $5$ is not one.

The four generators of $\mathbb F_{11}^*$ are: $2,6,7, 8\tag*{}$

As suggested in the comments by user Jyrki Lahtonen:
In case, $p$ is prime such that $(p-1)/2$ is odd i.e., $p\equiv 3\pmod{4}$ and $x$ is a generator of $\mathbb F_p^*$ then $-x$ is not a generator.

Nothing special
  • 3,690
  • 1
  • 7
  • 27
  • @xxxxxxxxx You are right... gcd(3,10)=1... I am deleting this – Nothing special Dec 10 '23 at 07:34
  • @xxxxxxxxx I have updated the answer. Can you reevaluate it, please? – Nothing special Dec 10 '23 at 07:44
  • @xxxxxxxxx I can eliminate only the perfect squares (e.g. $3^2$, $2^2$),and perfect 5-th powers e.g. $(-1)^5$ because they can't be generator, given that order of $\mathbb F_{11}= 10=2\times 5$. Is this correct? – Nothing special Dec 10 '23 at 07:54
  • 1
    The last claim is not right. Consider the field $\Bbb{F}_5$. The element $2$ is easily seen to have order four, hence a generator. But $3=-2$ is another generator. The $x$ vs. $-x$ game you want to play works as you hoped for only when we are looking for a primitive root modulo a prime $p$ such that $p\equiv3\pmod4$. In other words, the order of the group is even but not divisible by four. The difference comes from the fact that when $p-1$ is divisible by four, $(p-1)/2$ is still even. So when $x^{(p-1)/2}\equiv-1$ we also have $(-x)^{(p-1)/2}\equiv-1$. – Jyrki Lahtonen Dec 10 '23 at 08:49
  • Anyway, if $a=x^2$, $x\neq0$, then $a^5=x^{10}=1$ by Little Fermat. So it may be simpler to toss out all the quadratic residues. When the modulus is the larger part of a Sophie Germain pair of primes (like here $5$ and $11=2\cdot5+1$ are both primes), then it turns out that you only need toss out $-1$ and the quadratic residues. That's what the threads I linked to are all about. Observe also that the larger prime in an S.G. pair is always $\equiv3\pmod4$. Implying that $-1$ is not a quadratic residue. This plays nicely together with your game, as it follows that exactly one of $x,-x$ is a QR. – Jyrki Lahtonen Dec 10 '23 at 08:54
  • (cont'd) For the previous comment to make sense to you, you need to be familiar with basic properties of the quadratic residues. If you haven't seen that in your studies yet, ignore it :-) – Jyrki Lahtonen Dec 10 '23 at 08:55
  • @JyrkiLahtonen You are right... In this case, the fact that $\text{ord}(\mathbb F_{11})/2=5$ is odd plays an important role... I could conclude that $(-x)^5=-x$. I have removed the claim from my post... I am adding it here so that people who read the comment in future get the context: If the order of the group of units is even and $x$ is generator, $-x$ is not a generator. – Nothing special Dec 10 '23 at 08:56
1

For smaller primes, I usually follow one trick. If $a$ has to be a primitive root in $\mathbb{Z}_p$, then $a^{\frac{p-1}{2}}\equiv -1 \pmod p$( In other word $a$ is a quadratic non-residue modulo $p$. Need not worry if you aren't introduced to quadratic residues). There are always exactly $\frac{p-1}{2}$ elements in $\mathbb{Z}_p^{\star}$ which satisfy above congruence.

Further, there shall be $\varphi(p-1)$ primitive roots modulo $p$. These must be from these $\frac{p-1}{2}$ elements. Using these two, we can easily reduce the set of numbers for which the verification is needed. For instance:

In case of $p =11$, $\frac{p-1}{2}=5$ and there are exactly 5 elements in $\mathbb{Z}_{11}^\star$ that satisfy $a^5\equiv -1 \pmod {11}$. A primitive root must be one of these $5$ elements. As there are $\phi(10)=4$ primitive roots, $4$ out of the $5$ elements which satisfy above congruence must be primitive roots. Notice that $10$ which is $-1$ modulo $11$ is one element which satisfies above congruence. This is clearly not a primitive root as $o(10)=2$. Hence the primitive roots are all the remaining elements that satisfy $a^5\equiv 1 \pmod {11}$.

Quite clearly $2^5\equiv -1 \pmod {11}$ and hence it must be a primitive root modulo $11$. Once we get one primitive root, others can be easily found out, as others will be $2^3, 2^7, 2^9$ modulo $(11)$.

J. W. Tanner
  • 63,683
  • 4
  • 43
  • 88
Yathi
  • 2,859
0

Pick a random member of the group, and test whether it is a primitive root. Repeat until you find one that is.

To test whether $a$ is a primitive root modulo $p$, test whether

$$a^{(p-1)/d} \equiv 1 \pmod p,$$

for each prime divisor $d$ of $p-1$ (such that $1<d<p-1$). If this equivalence holds for any $d$, then $a$ is not a primitive root. If it fails to hold for all $d$, then $a$ is a primitive root.

See https://en.wikipedia.org/wiki/Primitive_root_modulo_n#Finding_primitive_roots.

D.W.
  • 5,958
0

Generally speaking, there is no simple algorithm to find primitive root.

But in this particular case, it is easy: Each element of $a\in\mathbb F_{11}^\times$ has order $1, 2, 5$ or $10$. We already know $-1=10$ is the only element of order $2$ (in the multiplicative group of any field). Now pick a random element $a\in \mathbb F_{11}\setminus\{0, \pm 1\}$, if $a$ has order $10$, we're done, otherwise $a$ has order $5$, then $-a$ must have order $10$. Only one element needs to be tested.

For simplicity, pick $a=2$, $2^5=32=-1$, we're done.

Just a user
  • 22,048