2

I am trying to work out exercise 61 from Rotman's Galois Theory, second edition:

61. Give the addition and multiplication tables of a field having eight elements. (Hint: Factor $x^8 -x$ over $\mathbb{Z}_2$.)

I am able to work out the previous exercise, which asks for a field with four elements by adjoining a suitable root of $x^4 - x$ to $\mathbb{Z}_2$. I suspect a similar strategy should be used for this exercise, but I haven't succeeded in working it out.

I understand that this field may be expressed as $\mathbb{Z}_2[x]/(p(x))$, where $p(x)$ is an irreducible polynomial of degree 3. This would allow me to work out the multiplication and addition tables. However, I don't think this is intended here, since the hint is not used.

When I try to use the hint and factor $x^8 - x$, I get as far as $$ x^8 - x = x(x - 1)(x^6 + x^5 + x^4 +x^3 +x^2 + x + 1) $$

I know that when I adjoin a root $\xi$ with $\xi^7 = -1$ I should be able to factor $$ x^6 + x^5 + x^4 +x^3 +x^2 + x + 1 = (x-\xi)(x-\xi^2)(x-\xi^3)(x-\xi^4)(x-\xi^5)(x-\xi^6) $$

But I cannot justify this with a calculation (when I try to apply polynomial factoring, the equations get quite involved and don't work out). Moreover, the field $\mathbb{Z}_2 / (x^6+x^5+x^4+x^3+x^2+x+1)$ resulting from adjoining this root has $2^6 = 64$ elements, not 8.

I think the proof of the following theorem should be used (quoted verbatim from the textbook):

Theorem 33 (Galois). For every prime $p$ and every positive integer $n$, there exists a field having exactly $p^n$ elements.

Proof. If there were a field $k$ with $|K| = p^n =q$, then $K^\# = K - \{ 0 \}$ would be a multiplicative group of order $q - 1$; by Lagrange's theorem, $a^{q - 1} = 1$ for all $a \in K^\#$. It follows that every element of $K$ would be a root of the polynomial $$ g(x) = x^q - x .$$ We can now begin the construction. By Kronecker's theorem, there is a field $E$ containing $\mathbb{Z}_p$ over which $g(x)$ splits. Define $F = \{ \alpha \in E : g(\alpha) = 0 \}$; that is, the set of all the roots of $g(x)$. Since the derivative $g'(x) = qx^{q - 1} - 1 = -1$ (because $q = p^n$ and $E$ has characteristic $p$), Lemma 32 shows that the $\gcd(g, g') = 1$, and so $g(x)$ has no repeated roots; that is, $|F| = q = p^n$.

We claim that $F$ is a field, which will complete the proof. If $a, b \in F$, then $a^q = a$ and $b^q = b$. Therefore, $(ab)^q = a^qb^q = ab$, and $ab \in F$. By Lemma 32(iii), replacing $b$ by $-b$, we have $(a - b)^q = a^q - b^q = a - b$, so that $a - b \in F$. Finally, if $a \not= 0$, then $a^{q-1} = 1$ so that $a^{-1} = a^{q - 2} \in F$ (because $F$ is closed under multiplication). $\ \bullet$

This theorem actually shows that the roots of $x^8 - x$ form a field when working in $\mathbb{Z}_2$. I understand the multiplicative structure of these roots, but I haven't succeeded in working out the splitting fields, and I'm having trouble working out the additive structure. I also haven't used yet that we are working in $\mathbb{Z}_2$.

Ruben
  • 804
  • 3
    Hint: Over $\Bbb{Z}_2$ you have the factorization $$x^8-x=x(x+1)(x^3+x+1)(x^3+x^2+1).$$ Show that those cubics are irreducible and let $\xi$ be a zero of one of them. – Jyrki Lahtonen Jul 01 '20 at 10:16
  • 2
    I prepared this post for referrals like this. It may also help. I constructed the field using the factor $x^3+x+1$. Observe that if $\xi$ is a zero of that factor, then $1/\xi$ is a zero of the other cubic. You can use whichever cubic you want, and the end results will be isomorphic (even though the details may look a bit different). – Jyrki Lahtonen Jul 01 '20 at 10:20
  • 1
    The main thing is that the elements of the field look like $b_0+b_1\xi+b_2\xi^2$, with $b_i\in{0,1}$ and the multiplication table determined by the cubic satisfied by $\xi$. The table in my post lists these elements also as $0,1,\xi,\xi^2,...,\xi^6,\xi^7=1$. The former is a better way for handling addition, the latter for handling multiplication. The point of the table is for converting from one representation to the other all according to what you want to do. It may be a bit early for you to look into that seriously, but it does help in getting familiarized with this theme. – Jyrki Lahtonen Jul 01 '20 at 10:27
  • @JyrkiLahtonen That clears it up completely for me! I wasn't able to find a factorization of $x^6+x^5+x^4+x^3+x^2+x+1$, so I assumed it was irreducible (if I'm not mistaken, it is irreducible over $\mathbb{Q}$, but not over $\mathbb{Z}_2$). Your answer on this question helps me with that :) – Ruben Jul 01 '20 at 11:44
  • 1
    Glad to hear that. If you want to, you can write up an answer to your question also. That way you get more feedback. Your call, of course. – Jyrki Lahtonen Jul 01 '20 at 13:26

1 Answers1

2

Based on Jyrki Lahtonen's comment I am able to work out the answer.

My mistake was that I didn't see an obvious factorization of $p(x) = x^6+x^5+x^4+x^3+x^2+x+1$. Polynomials of this kind (cyclotomic polynomials) are irreducible in $\mathbb{Q}[x]$ -- but not necessarily in $\mathbb{Z}_2$!

Indeed we can just try dividing by some different polynomials as suggest in the comments and answers on this question. Since $p(x)$ is of degree 6, it must either be prime or have a divisor of degree three or smaller. We can also observe that any divisor must have 1 as the constant term.

Trying to divide $p(x)$ by $x + 1$, $x^2 + 1$, $x^2 + x + 1$, $x^3 + 1$ does not result into a factorization. However, when we try to divide by $x^3 + x + 1$ (and not forgetting to do arithmetic modulo 2!), we obtain $$ x^8 - x = x(x - 1)(x^3 + x + 1)(x^3 + x^2 + 1) $$

Now we adjoin a root $\xi$ of $x^3 + x + 1$. The field that is formed this way is isomorphic to $\mathbb{Z}_2/(x^3 + x + 1)$. That is, our field is isomorphic to polynomials of order < 3 (since we have $\xi^3 = \xi + 1$) with coefficients in $\mathbb{Z}_2$. So we have the elements $0, 1, \xi, \xi^2, \xi^3 = \xi + 1, \xi^4 = \xi^2 + \xi, \xi^5 = \xi^2 + \xi + 1, \xi^6 = \xi^2 + 1$.

Now working out the multiplication and addition tables is trivial. As Jyrki Lahtonen has noted, it is easier to use the form $\xi^k$ for multiplication, and to use the form $a \xi^2 + b \xi + c$ for addition. Of course, we have $\xi^a + \xi^b = \xi^{a + b}$ (where $a + b$ might need to be reduced modulo 7, since $\xi ^ 7 = 1$).

The addition table is as follows (I left out the zero polynomial to keep the size of the table manageable):

$$\begin{matrix} + & 1 & \xi & \xi^2 & \xi + 1 & \xi^2 + \xi & \xi^2 + \xi + 1 & \xi^2 + 1\\ 1 & 0 & \xi + 1 & \xi^2 + 1 & \xi & \xi^2 + \xi + 1 & \xi^2 + \xi & \xi^2 \\ \xi & \xi + 1 & 0 & \xi^2 + \xi & 1 & \xi^2 & \xi^2 + 1 & \xi^2 + \xi + 1 \\ \xi^2 & \xi^2 + 1 & \xi^2 + \xi & 0 & \xi^2 + \xi + 1 & \xi & \xi + 1 & 1 \\ \xi + 1 & \xi & 1 & \xi^2 + \xi + 1 & 0 & \xi^2 + 1 & \xi^2 & \xi^2 + \xi \\ \xi^2 + \xi & \xi^2 + \xi + 1 & \xi^2 & \xi & \xi^2 + 1 & 0 & 1 & \xi + 1 \\ \xi^2 + \xi + 1 & \xi^2 + \xi & \xi^2 + 1 & \xi + 1 & \xi^2 & 1 & 0 & \xi \\ \xi^2 + 1 & \xi^2 & \xi^2 + \xi + 1 & 1 & \xi^2 + \xi & \xi + 1 & \xi & 0 \\ \end{matrix}$$

Ruben
  • 804