Integral Domains and Fields

In this lesson you’ll learn what an integral domain is, why zero divisors break cancellation, exactly when $\mathbb{Z}_n$ is a field, and the theorem that every finite integral domain is automatically a field.

An integral domain is a commutative ring with unity $1 \ne 0$ and no zero divisors:

$$ab = 0 \implies a = 0 \text{ or } b = 0$$

The name comes from $\mathbb{Z}$, the prototype. The condition is exactly what you assume every time you solve an equation by factoring: from $(x-2)(x-3) = 0$ you conclude $x = 2$ or $x = 3$, and that step is only valid in an integral domain.

In an integral domain, if $a \ne 0$ and $ab = ac$, then $b = c$.

Proof: $ab - ac = 0$, so $a(b - c) = 0$. Since $a \ne 0$ and there are no zero divisors, $b - c = 0$. ∎

This is the practical payoff. Cancellation is so familiar that its failure is genuinely disorienting.

×	0	1	2	3	4	5
0	0	0	0	0	0	0
1	0	1	2	3	4	5
2	0	2	4	0	2	4
3	0	3	0	3	0	3
4	0	4	2	0	4	2
5	0	5	4	3	2	1

Zeros appear off the top row and left column: 2 × 3 = 0 and 3 × 4 = 0. Row 2 reads 0, 2, 4, 0, 2, 4, repeating, so 2x = 2y does not force x = y.

Identity: 1. Cells holding it mark inverse pairs.

Look at row 2: it reads $0, 2, 4, 0, 2, 4$. The value 2 appears twice, at $2 \cdot 1$ and $2 \cdot 4$. So $2 \cdot 1 = 2 \cdot 4$ while $1 \ne 4$: cancellation fails outright.

Compare a prime modulus.

×	0	1	2	3	4
0	0	0	0	0	0
1	0	1	2	3	4
2	0	2	4	1	3
3	0	3	1	4	2
4	0	4	3	2	1

The only zeros sit in row 0 and column 0. Every other row is a rearrangement of 1, 2, 3, 4, so every nonzero element has an inverse.

Identity: 1. Cells holding it mark inverse pairs.

No stray zeros, and every nonzero row contains a 1. That’s a field.

A field is a commutative ring with unity in which every nonzero element has a multiplicative inverse.

Equivalently: $(F, +)$ is an abelian group, $(F \setminus \{0\}, \times)$ is an abelian group, and multiplication distributes over addition. Two groups sharing a set, glued by distributivity.

Every field is an integral domain. If $ab = 0$ with $a \ne 0$, multiply by $a^{-1}$:

$$b = a^{-1}(ab) = a^{-1} \cdot 0 = 0$$

So no zero divisors exist. The converse fails: $\mathbb{Z}$ is an integral domain and not a field, since 2 has no integer reciprocal.

$\mathbb{Z}_n$ is a field $\iff$ $n$ is prime.

Both directions are short.

If $n$ is composite, write $n = ab$ with $1 < a, b < n$. Then $a$ and $b$ are nonzero in $\mathbb{Z}_n$ but $ab = n \equiv 0$. Zero divisors exist, so it isn’t even an integral domain.

If $n = p$ is prime and $a \not\equiv 0$, then $\gcd(a,p) = 1$, so Bézout gives integers $x, y$ with $ax + py = 1$. Reducing mod $p$: $ax \equiv 1$, so $x = a^{-1}$.

Because these fields are so important they get a name: $\mathbb{F}_p$, or $GF(p)$ for “Galois field.”

Here’s a result with no analogue for infinite rings.

A finite integral domain is a field.

Proof: let $D$ be a finite integral domain and $a \ne 0$. Consider the map $x \mapsto ax$ from $D$ to $D$. It’s injective, because $ax = ay$ implies $x = y$ by cancellation. An injective map from a finite set to itself is surjective, so something maps to 1: there is an $x$ with $ax = 1$. ∎

So $a$ is invertible. Finiteness is what forces it - the same argument fails for $\mathbb{Z}$, where $x \mapsto 2x$ is injective but misses 1 because the set is infinite.

This is why the distinction between domain and field, which matters a great deal for $\mathbb{Z}$ and $\mathbb{R}[x]$, evaporates in the finite world.

The characteristic of a ring with unity is the smallest positive $n$ with

$$\underbrace{1 + 1 + \dots + 1}_{n \text{ times}} = 0$$

and $0$ if no such $n$ exists.

$\mathbb{Z}$, $\mathbb{Q}$, $\mathbb{R}$, $\mathbb{C}$ all have characteristic 0. $\mathbb{Z}_n$ has characteristic $n$.

The characteristic of an integral domain is 0 or prime. Suppose it were composite, $n = ab$ with $1 < a,b < n$. Then in the ring, $(a \cdot 1)(b \cdot 1) = n \cdot 1 = 0$, and neither factor is 0 because $n$ is the smallest such value. That’s a zero divisor, contradiction. ∎

In characteristic $p$, the identity $(x+y)^p = x^p + y^p$ holds, because all the middle binomial coefficients are divisible by $p$. Sometimes called the “freshman’s dream,” it’s false over $\mathbb{R}$ and true in characteristic $p$, and it underlies a good deal of finite field theory.

Example 1: Is $\mathbb{Z}_{15}$ an integral domain?

Solution. $15 = 3 \cdot 5$, so $3$ and $5$ are nonzero in $\mathbb{Z}_{15}$ but $3 \cdot 5 = 15 \equiv 0$.

No. Zero divisors exist, so not a domain and certainly not a field.

The units are $U(15) = \{1,2,4,7,8,11,13,14\}$, of size $\varphi(15) = 8$, and the other six nonzero elements are zero divisors.

Example 2: Solve $x^2 = 1$ in $\mathbb{Z}_8$.

Solution. Test all eight:

$$0^2=0,\ 1^2=1,\ 2^2=4,\ 3^2=1,\ 4^2=0,\ 5^2=1,\ 6^2=4,\ 7^2=1$$

Solutions: $x \in \{1, 3, 5, 7\}$ - four of them.

Over a field, $x^2 = 1$ factors as $(x-1)(x+1) = 0$ and has at most two solutions. Here $\mathbb{Z}_8$ has zero divisors, the factoring argument fails, and a quadratic gets four roots.

A polynomial of degree $n$ over a field has at most $n$ roots. That familiar fact needs the field hypothesis, and this example is what goes wrong without it.

Example 3: Find $7^{-1}$ in $\mathbb{Z}_{11}$.

Solution. $11$ is prime, so the inverse exists. Look for $x$ with $7x \equiv 1 \pmod{11}$.

$$7 \cdot 1 = 7, \quad 7 \cdot 2 = 14 \equiv 3, \quad 7 \cdot 3 = 21 \equiv 10, \quad 7 \cdot 4 = 28 \equiv 6, \quad 7 \cdot 5 = 35 \equiv 2, \quad 7 \cdot 8 = 56 \equiv 1$$ $$7^{-1} = 8$$

For larger moduli, use the extended Euclidean algorithm rather than trial: it finds $x, y$ with $7x + 11y = 1$ in a handful of steps regardless of size.

Example 4: Is $\mathbb{Z}[x]$ an integral domain?

Solution. Commutative ✓, unity is the constant polynomial 1 ✓.

Zero divisors: if $f$ and $g$ are nonzero with leading terms $a_mx^m$ and $b_nx^n$, then $fg$ has leading term $a_mb_nx^{m+n}$. Since $\mathbb{Z}$ has no zero divisors, $a_mb_n \ne 0$, so $fg \ne 0$.

Yes, an integral domain. Not a field, since $x$ has no inverse: any $fg = 1$ forces $\deg f + \deg g = 0$, so both are constants, and $x$ is not.

Generally, $R[x]$ is an integral domain exactly when $R$ is. The degree argument above is where the hypothesis gets used.

Example 5: A finite domain must be a field.

Verify the theorem on $\mathbb{Z}_7$ by exhibiting the inverses.

Solution. $\mathbb{Z}_7$ is a finite integral domain, so the theorem promises a field.

$$1^{-1}=1, \quad 2^{-1}=4, \quad 3^{-1}=5, \quad 4^{-1}=2, \quad 5^{-1}=3, \quad 6^{-1}=6$$

Check a couple: $2 \cdot 4 = 8 \equiv 1$ ✓, $3 \cdot 5 = 15 \equiv 1$ ✓.

Every nonzero element inverted, as promised.

Example 6: Characteristic of a product.

Find the characteristic of $\mathbb{Z}_4 \times \mathbb{Z}_6$.

Solution. The unity is $(1,1)$. Adding it to itself $n$ times gives $(n \bmod 4, n \bmod 6)$, which is $(0,0)$ exactly when $4 \mid n$ and $6 \mid n$.

$$\text{char} = \operatorname{lcm}(4,6) = 12$$

Since 12 is composite, this ring cannot be an integral domain - and indeed $(2,0)(0,3) = (0,0)$ with both factors nonzero.

Example 7: The freshman’s dream.

Verify $(x+y)^3 = x^3 + y^3$ in $\mathbb{Z}_3$.

Solution. Expand normally:

$$(x+y)^3 = x^3 + 3x^2y + 3xy^2 + y^3$$

In characteristic 3, $3 = 0$, so both middle terms vanish:

$$(x+y)^3 = x^3 + y^3$$

Check with $x = y = 1$: left side $(1+1)^3 = 2^3 = 8 \equiv 2$; right side $1 + 1 = 2$ ✓.

This map $x \mapsto x^p$ is the Frobenius endomorphism, and it is the engine of finite field theory.

Cryptography. Elliptic-curve cryptography works over $\mathbb{F}_p$ for a large prime $p$, and needing a field rather than just a ring is essential: the group law on the curve involves dividing by coordinate differences. Over $\mathbb{Z}_n$ with composite $n$ that division can fail, and the failure leaks a factor of $n$ - the basis of Lenstra’s elliptic-curve factoring method.

Reed–Solomon codes. QR codes, CDs, DVDs and spacecraft telemetry use codes over finite fields. Decoding solves linear systems, which requires inverting elements, which requires a field. The field is usually $GF(256)$, so one field element is exactly one byte.

Hash tables and pseudorandom generators. Choosing a prime modulus makes the arithmetic happen in a field, so every nonzero multiplier is invertible and no probe sequence gets trapped in a short cycle. Composite moduli create zero divisors and, with them, degenerate behaviour.

Computer algebra. Whether an algorithm applies depends on the ring type: Gaussian elimination needs a field, the Euclidean algorithm needs a Euclidean domain, factoring behaves differently over $\mathbb{Z}$ and $\mathbb{Q}$. Systems like SageMath dispatch on exactly the distinctions in this lesson.

Fixed-width integer bugs. 32-bit arithmetic is $\mathbb{Z}_{2^{32}}$, which is riddled with zero divisors. Code assuming that a * b == a * c implies b == c is assuming an integral domain it does not have, and this is a genuine source of security vulnerabilities.