Multiply $3 + 2i$ by $1 - 4i$ by the usual rules: expand, and replace $i^2$ by $-1$.
$$(3 + 2i)(1 - 4i) = 3 - 12i + 2i - 8i^2 = 11 - 10i.$$
Now multiply the polynomials $3 + 2x$ and $1 - 4x$ in $\mathbb{R}[x]$ and divide by $x^2 + 1$:
$$(3 + 2x)(1 - 4x) = -8x^2 - 10x + 3 = (x^2 + 1)(-8) + (11 - 10x).$$
The remainder is the complex product with $x$ in place of $i$. Why should a remainder on division by $x^2 + 1$ know how to multiply complex numbers?
It is no coincidence. Replacing $i^2$ by $-1$ is the same as discarding multiples of $x^2 + 1$, which is congruence modulo $x^2 + 1$. So the complex numbers can be described as the congruence classes modulo $x^2 + 1$, added and multiplied through representatives. In general, the classes modulo $p(x)$ form a ring, written $F[x]/(p(x))$. The same construction with $x^2 - 2$ over the rationals produces the numbers $a + b\sqrt{2}$. With other moduli it produces number systems that were not known in advance, including finite fields.
This article follows the integers modulo $n$ step by step. First we prove that the operations on classes are well defined and obey the ring laws. Then comes what is new: a copy of $F$ and the class of $x$, which satisfies the modulus as an equation and identifies two of these rings with $\mathbb{C}$ and $\mathbb{Q}(\sqrt 2)$. Finally, Bezout's identity decides which classes have inverses, as it did for units in $\mathbb{Z}_n$, and a reducible modulus produces zero divisors.
The article assumes the previous one, Congruence Modulo a Polynomial; the running quartic $f(x) = (x^2 - 2)(x^2 + x + 1)$ returns as a modulus. This is the second article of Congruence in Polynomial Rings and Field Extensions in the MTH 532 notes. It covers §5.2 of the course text (Hungerford, 2014). Throughout, $F$ is a field and $p(x) \in F[x]$ is a polynomial of degree $d \ge 1$.
The set $F[x]/(p(x))$
Definition. $F[x]/(p(x))$ is the set of congruence classes of $F[x]$ modulo $p(x)$.
The notation, read "$F[x]$ modulo $p(x)$", says how the set was made, as $\mathbb{Z}/n\mathbb{Z}$ does for the integers modulo $n$; the parentheses around $p(x)$ stand for the set of its multiples. By the theorem on classes, each class contains exactly one polynomial that is 0 or of degree less than $d$, so
$$F[x]/(p(x)) = \{\, [a_0 + a_1 x + \cdots + a_{d-1}x^{d-1}] : a_0, \ldots, a_{d-1} \in F \,\},$$
with different coefficient lists giving different classes. Over a field with $q$ elements the set has $q^d$ elements; over $\mathbb{Q}$, $\mathbb{R}$ or $\mathbb{C}$ it is infinite. As with $\mathbb{Z}_n$, the elements are sets of polynomials, and $[x^2] = [-1]$ in $\mathbb{R}[x]/(x^2 + 1)$ is a true equation between sets, though $x^2 \neq -1$ as polynomials.
Adding and multiplying classes
To compute with classes we need operations on them. The definitions are the ones used for integers:
$$[f(x)] + [g(x)] = [f(x) + g(x)], \qquad [f(x)] \cdot [g(x)] = [f(x)g(x)].$$
Each says: choose a representative of each class, combine the representatives in $F[x]$, and take the class of the result. The choice must not matter. In the opening example, $[3 + 2x]$ is also the class of $x^3 + 3x + 3$ (add $x(x^2 + 1)$), and $[1 - 4x]$ is also the class of $-5x^2 - 4x - 4$ (subtract $5(x^2 + 1)$). The product of these representatives is the degree 5 polynomial
$$-5x^5 - 4x^4 - 19x^3 - 27x^2 - 24x - 12,$$
and the script finds its remainder on division by $x^2 + 1$ to be $11 - 10x$ again.
The operations are well defined
The idea: two representatives of one class are congruent, and congruences survive sums and products.
Theorem. If $[f] = [f']$ and $[g] = [g']$ in $F[x]/(p(x))$, then $[f + g] = [f' + g']$ and $[fg] = [f'g']$. Consequently the rules above define an addition and a multiplication on $F[x]/(p(x))$.
Proof. Equal classes have congruent representatives, so $f \equiv f'$ and $g \equiv g' \pmod p$. By the theorem on sums and products of congruences in Congruence Modulo a Polynomial, $f + g \equiv f' + g'$ and $fg \equiv f'g'$, and congruent polynomials have equal classes. $\blacksquare$
This is Hungerford's Theorem 5.6, and it is the integer proof with the integer theorem on congruences replaced by its polynomial twin. It has the same practical meaning: arithmetic in $F[x]/(p(x))$ can be carried out on the remainders, reducing modulo $p(x)$ after each step.
Rules that are not well defined
Several natural operations on polynomials ignore this test and fail it. Modulo $x^2 + 1$ in $\mathbb{R}[x]$, the polynomials $x^2 + 1$ and $0$ lie in the same class, and so do $x^2$ and $-1$.
- Evaluation at a number. The rule $[h] \mapsto h(0)$ assigns the value $1$ to the class of $x^2 + 1$ and the value $0$ to the class of $0$, which is the same class. Evaluation at $1$ fails the same way on $x^2$ and $-1$.
- The derivative. The rule $[h] \mapsto [h']$ sends $[x^2 + 1]$ to $[2x]$ and $[0]$ to $[0]$, and $2x \not\equiv 0$.
- The degree. The class of $x^2$ contains polynomials of every degree from 0 (namely $-1$) upward, so "the degree of a class" means nothing, although the degree of its standard representative does.
Each rule fails because it does not ignore multiples of $p(x)$.
Check your understanding. Evaluation at 0 and at 1 failed modulo $x^2 + 1$. For which numbers $c$, real or complex, is the rule $[h] \mapsto h(c)$ well defined?
Exactly for the roots $c = \pm i$. If $p(c) = 0$, every multiple $k(x)p(x)$ has the value $k(c) \cdot 0 = 0$, so all representatives of a class agree at $c$. If $p(c) \neq 0$, then $p$ and $0$ lie in one class but take different values. No real number qualifies. That observation is the source of the two isomorphisms later in this article.
The laws of arithmetic
Which laws do the operations obey? The idea: sending a polynomial to its class turns sums into sums and products into products, so every law of $F[x]$ carries over.
Theorem. $F[x]/(p(x))$, with the operations above, is a commutative ring with identity $[1]$, and $[1] \neq [0]$.
Proof. Closure holds because each operation produces a class. For the other laws, let $\pi(h) = [h]$ be the map from $F[x]$ to $F[x]/(p(x))$ that sends each polynomial to its class. The definitions of the operations, now known to be well defined, say exactly that
$$\pi(f + g) = \pi(f) + \pi(g), \qquad \pi(fg) = \pi(f)\,\pi(g).$$
Every law of a commutative ring with identity is an equation between expressions built from elements with sums and products, such as $[f]([g] + [h]) = [f][g] + [f][h]$. Applying the two rules from the inside out turns each side into $\pi$ of the corresponding expression in $F[x]$, here $\pi(f(g + h))$ and $\pi(fg + fh)$. These are equal, because the law holds in $F[x]$, which is a commutative ring with identity. The zero, the identity and the negatives are $[0]$, $[1]$ and $[-f]$, because $f + 0 = f$, $f \cdot 1 = f$ and $f + (-f) = 0$ in $F[x]$. Finally, $1$ and $0$ are different polynomials of degree less than $d$ (here $d \ge 1$ is used), so their classes are different. $\blacksquare$
The argument is the one given for $\mathbb{Z}_n$ in The Integers Modulo n: any identity among sums and products that holds in $F[x]$ holds among the classes. The theorem is the first half of Hungerford's Theorem 5.7. For the two sixteen-element rings in the last section, the script also checks commutativity, associativity and distributivity on every pair and on all 4,096 triples of elements.
A copy of $F$, and the class of $x$
Does the new ring still contain $F$, and what does it add? The constants of $F$ sit inside $F[x]$, and their classes behave exactly like the constants. The idea: two different constants differ by a nonzero constant, too small to be a multiple of $p$.
Proposition. The map $\iota(c) = [c]$ from $F$ to $F[x]/(p(x))$ is one-to-one and satisfies $\iota(a + b) = \iota(a) + \iota(b)$ and $\iota(ab) = \iota(a)\,\iota(b)$. Its image $\{[c] : c \in F\}$ is a subring isomorphic to $F$.
Proof. The two equations are the definitions of the operations applied to constant polynomials. If $[a] = [b]$, then $p \mid a - b$, and $a - b$ is a constant, so it is 0 or has degree 0; a nonzero polynomial divisible by $p$ has degree at least $d \ge 1$. So $a - b = 0$. The image contains $[0]$ and is closed under subtraction and multiplication by the two equations, so it is a subring, and $\iota$ is a bijection onto it that respects both operations. $\blacksquare$
This is the second half of Hungerford's Theorem 5.7. It allows a convention like the one for $\mathbb{Z}_n$: write $c$ for the class $[c]$, and regard $F$ as a subset of $F[x]/(p(x))$. In $\mathbb{Z}_3[x]/(p(x))$, for instance, $[2] + [2] = [1]$ and $[2][2] = [1]$, just as $2 + 2 = 1$ and $2 \cdot 2 = 1$ in $\mathbb{Z}_3$, so nothing is lost by dropping the brackets. With this convention $F[x]/(p(x))$ is a ring containing $F$ (Hungerford's Theorem 5.8).
The class of $x$ is the one genuinely new element, and it pays to give it a name. Let
$$\alpha = [x].$$
By the definition of the operations, $[a_0 + a_1 x + \cdots + a_k x^k] = a_0 + a_1\alpha + \cdots + a_k\alpha^k$, so every element of $F[x]/(p(x))$ is a polynomial expression in $\alpha$ with coefficients in $F$, and the standard representatives give each element exactly one expression
$$a_0 + a_1\alpha + \cdots + a_{d-1}\alpha^{d-1}, \qquad a_i \in F.$$
The same calculation applied to the modulus itself gives
$$p(\alpha) = [p(x)] = [0] = 0,$$
since $p(x) \equiv 0 \pmod{p(x)}$. So $\alpha$ is a root of $p$ in the new ring, whether or not $p$ has a root in $F$. That one line is the heart of the third article, where it becomes Kronecker's theorem. For computing it gives the reduction rule: if $p$ is monic, $\alpha^d = -(c_{d-1}\alpha^{d-1} + \cdots + c_0)$, the rule "replace $x^2$ by 2" of the previous article now written as an equation.
Two familiar fields
$\mathbb{R}[x]/(x^2 + 1)$ and the complex numbers
Here $\alpha^2 = -1$, and the elements are $a + b\alpha$ with $a, b \in \mathbb{R}$. Multiplying and replacing $\alpha^2$,
$$(a + b\alpha)(c + d\alpha) = ac + (ad + bc)\alpha + bd\alpha^2 = (ac - bd) + (ad + bc)\alpha,$$
which is the rule for multiplying complex numbers, with $\alpha$ in the role of $i$. Sums agree as well. So the map $a + b\alpha \mapsto a + bi$ is a bijection from $\mathbb{R}[x]/(x^2 + 1)$ onto $\mathbb{C}$ that respects both operations: the two rings are the same up to the names of their elements. Hungerford (2014, §5.3) and Judson (2026, Example 21.1.14) both make this identification, and the course text adds that one could take it as the definition of $\mathbb{C}$, which removes any doubt about whether a number whose square is $-1$ exists: the class of $x$ is one.
$\mathbb{Q}[x]/(x^2 - 2)$ and $\mathbb{Q}(\sqrt 2)$
Here $\alpha^2 = 2$, the elements are $a + b\alpha$ with $a, b \in \mathbb{Q}$, and
$$(a + b\alpha)(c + d\alpha) = (ac + 2bd) + (ad + bc)\alpha.$$
The opening computation of Congruence Modulo a Polynomial, $(1 + \alpha)^5 = 41 + 29\alpha$, took place in this ring. The comparison is now with the set of real numbers
$$\mathbb{Q}(\sqrt 2) = \{\, a + b\sqrt 2 : a, b \in \mathbb{Q} \,\}.$$
Both identifications are cases of one argument, based on the check above: evaluation at a root of the modulus is well defined, and it loses no information when no smaller polynomial vanishes at the root.
Proposition. Let $F$ be a subfield of a field $E$, and let $c \in E$ be a root of $p(x)$. Then
$$\varphi([h(x)]) = h(c)$$
is a well-defined map from $F[x]/(p(x))$ to $E$ that respects sums and products. If no nonzero polynomial over $F$ of degree less than $d$ has $c$ as a root, then $\varphi$ is one-to-one, and its image is $\{a_0 + a_1 c + \cdots + a_{d-1}c^{d-1} : a_i \in F\}$.
Proof. Well defined: if $[h] = [h']$, then $h' = h + kp$ for some $k$, and $h'(c) = h(c) + k(c)p(c) = h(c)$. Evaluation at $c$ respects sums and products of polynomials (evaluation needs only that $E$ is commutative), so $\varphi([f] + [g]) = \varphi([f + g]) = (f + g)(c) = f(c) + g(c)$, and likewise for products. For the second statement, write a class through its standard representative $r$, which is 0 or of degree less than $d$. If $\varphi([r]) = r(c) = 0$ with $r \neq 0$, then $r$ is a nonzero polynomial of degree less than $d$ with root $c$, which the hypothesis excludes. So $\varphi([r]) = 0$ only for $[r] = 0$, and if $\varphi([r]) = \varphi([s])$, then $\varphi([r - s]) = 0$ and $[r] = [s]$. The image consists of the values $r(c)$, which are the listed expressions. $\blacksquare$
For $\mathbb{R}[x]/(x^2 + 1)$, take $E = \mathbb{C}$ and $c = i$. A nonzero real polynomial of degree less than 2 is $a + bx$ with $a$ and $b$ not both 0, and $a + bi = 0$ forces $a = b = 0$; so $\varphi$ is one-to-one, and its image is all of $\mathbb{C}$.
For $\mathbb{Q}[x]/(x^2 - 2)$, take $E = \mathbb{R}$ and $c = \sqrt 2$. If $a + b\sqrt 2 = 0$ with $a, b$ rational and not both 0, then $b \neq 0$ and $\sqrt 2 = -a/b$ would be rational, which the rational root test rules out, since $x^2 - 2$ has no rational roots. So $\varphi$ is a bijection from $\mathbb{Q}[x]/(x^2 - 2)$ onto $\mathbb{Q}(\sqrt 2)$ that respects both operations. In the language of a later unit, $\varphi$ is a ring homomorphism, and both identifications are instances of the first isomorphism theorem (Judson, 2026, Proposition 21.1.12).
Under $\varphi$, the running example $f(x) = (x^2 - 2)(x^2 + x + 1)$ of the previous unit has the class $[f] = 0$ in $\mathbb{Q}[x]/(x^2 - 2)$, which says that $f(\sqrt 2) = 0$. In floating point the script gets about $1.3 \times 10^{-15}$, zero up to rounding.
Units
In $\mathbb{C}$ every nonzero element has an inverse. Is that so for every modulus? In $\mathbb{Z}_n$ the class $[a]$ has an inverse exactly when $\gcd(a, n) = 1$, by Bezout's identity. The same holds here, with the same idea: $hu + pv = 1$ says exactly that $[h][u] = [1]$.
Theorem. Let $h(x) \in F[x]$. The class $[h(x)]$ is a unit of $F[x]/(p(x))$ if and only if $\gcd(h(x), p(x)) = 1$.
Proof. If $\gcd(h, p) = 1$, Bezout's identity gives $u$ and $v$ with $hu + pv = 1$. Then $hu \equiv 1 \pmod p$, so $[h][u] = [1]$.
Conversely, if $[h][u] = [1]$, then $hu - 1 = pv$ for some $v$, that is, $hu - pv = 1$. Every common divisor of $h$ and $p$ divides $1$, so it is a nonzero constant, and the monic gcd is 1. $\blacksquare$
The first half is Hungerford's Theorem 5.9; the converse is the integer argument again. The proof finds the inverse: run the extended Euclidean algorithm on $h$ and $p$ and read off $u$.
In $\mathbb{Q}[x]/(x^3 - 2)$, where $\alpha = [x]$ satisfies $\alpha^3 = 2$, take $h = x + 1$. One division, $x^3 - 2 = (x + 1)(x^2 - x + 1) - 3$, leaves the nonzero constant remainder $-3$, so $\gcd(x + 1, x^3 - 2) = 1$ and
$$(x + 1)(x^2 - x + 1) - (x^3 - 2) = 3, \qquad \text{hence} \qquad (x + 1) \cdot \tfrac{1}{3}(x^2 - x + 1) \equiv 1 \pmod{x^3 - 2},$$
and $[x + 1]^{-1} = \tfrac{1}{3} - \tfrac{1}{3}\alpha + \tfrac{1}{3}\alpha^2$. The real cube root $c = \sqrt[3]{2}$ is a root of $x^3 - 2$, so evaluation at $c$ is well defined and respects products, by the first part of the proposition above. It carries the equation $[x + 1][u] = [1]$, with $u = \tfrac{1}{3}(x^2 - x + 1)$, to $(1 + c)\,u(c) = 1$, that is,
$$\frac{1}{1 + \sqrt[3]{2}} = \frac{1 - \sqrt[3]{2} + \sqrt[3]{4}}{3},$$
a rationalization that multiplying by a conjugate, the trick for square roots, does not reach. The script confirms both sides numerically, $0.4424933\ldots$ in each case.
In $\mathbb{Q}[x]/(f(x))$, with the running example itself as the modulus, the class of $x$ is a unit, since the irreducible $x$ does not divide $f$ (because $f(0) = -2 \neq 0$). Here the inverse can be read off directly: $f(x) = x(x^3 + x^2 - x - 2) - 2$, so $x \cdot \tfrac{1}{2}(x^3 + x^2 - x - 2) \equiv 1 \pmod f$.
Check your understanding. Without running the Euclidean algorithm, decide which of $[x - 1]$ and $[x^2 + x + 1]$ is a unit of $\mathbb{Q}[x]/(f(x))$.
$[x - 1]$ is a unit and $[x^2 + x + 1]$ is not. By the factor theorem, $x - 1$ divides $f$ only if $f(1) = 0$, and the rational root test found $f(1) = -3$; so $\gcd(x - 1, f) = 1$. But $x^2 + x + 1$ is a factor of $f$, so it is its own gcd with $f$. Only gcds were needed, not inverses.
Zero divisors and a reducible modulus
What happens to the classes that are not units? The class $[x^2 - 2]$ in $\mathbb{Q}[x]/(f(x))$ is not a unit, since $\gcd(x^2 - 2, f) = x^2 - 2$. It is something worse:
$$[x^2 - 2]\,[x^2 + x + 1] = [f(x)] = 0,$$
although neither factor is 0, because neither $x^2 - 2$ nor $x^2 + x + 1$ is divisible by $f$, which has larger degree. A factorization of the modulus becomes a pair of zero divisors, just as $4 \cdot 6 = 24$ became $[4][6] = 0$ in $\mathbb{Z}_{24}$. As for $\mathbb{Z}_n$, every nonzero class is one or the other.
Theorem. Every nonzero element of $F[x]/(p(x))$ is either a unit or a zero divisor, and not both.
The idea: a non-unit $h$ shares a factor $d$ with $p$, and multiplying $h$ by the rest of $p$ completes a multiple of $p$, as $x^2 + x + 1$ did for $x^2 - 2$ above.
Proof. Let $[h] \neq 0$ be a non-unit, so $d = \gcd(h, p) \neq 1$ by the previous theorem, and $\deg d \ge 1$. Write $p = dp'$ and $h = dh'$. Then
$$h\,p' = d h' p' = h'\,p \equiv 0 \pmod p,$$
so $[h][p'] = 0$. And $[p'] \neq 0$: $p'$ is nonzero with $\deg p' = \deg p - \deg d < \deg p$, so $p$ does not divide it. Hence $[h]$ is a zero divisor. A unit is never a zero divisor: if $[h][u] = 1$ and $[h][g] = 0$, then $[g] = [u][h][g] = 0$. $\blacksquare$
For a check, reducing the running example modulo 3 gives a ring small enough to search completely. In $\mathbb{Z}_3[x]$,
$$\bar f(x) = x^4 + x^3 + 2x^2 + x + 1 = (x + 2)^2 (x^2 + 1),$$
so $\mathbb{Z}_3[x]/(\bar f)$ has $3^4 = 81$ elements. The script sorts them: 48 units (the classes of polynomials sharing no factor with $\bar f$), 32 zero divisors, and the zero class, with no nonzero element in both groups or in neither. For example $[x + 2]\,[(x + 2)(x^2 + 1)] = 0$, where $(x + 2)(x^2 + 1) = x^3 + 2x^2 + x + 2$, and $[x]$ is a unit with inverse $[2x^3 + 2x^2 + x + 2]$. The count of units mirrors Euler's $\varphi$ function: $81 \cdot (1 - \tfrac{1}{3})(1 - \tfrac{1}{9}) = 48$, where $3$ and $9$ are the sizes of $\mathbb{Z}_3[x]$ modulo the distinct irreducible factors $x + 2$ and $x^2 + 1$. The reason is the same as for $\varphi$, the Chinese remainder theorem, whose polynomial form splits this ring into pieces modulo $(x + 2)^2$ and $x^2 + 1$.
Two rings with sixteen elements
How much does factoring matter when everything else is equal? Over $\mathbb{Z}_2$, every quartic modulus gives a ring with $2^4 = 16$ elements, and whether the modulus factors changes everything. For two of them, the gcd theorem predicts the units before any product is computed.
The modulus $x^4 + 1$ is as reducible as a quartic can be: over $\mathbb{Z}_2$, $(x + 1)^2 = x^2 + 1$, so $(x + 1)^4 = (x^2 + 1)^2 = x^4 + 1$. A class $[h]$ is therefore a unit exactly when $x + 1$ does not divide $h$, that is, by the factor theorem, when $h(1) = 1$, which says that $h$ has an odd number of terms. That predicts 8 units and 7 zero divisors, and the count mirrors Euler's $\varphi$ again: $16 \cdot (1 - \tfrac{1}{2}) = 8$, with 2 the size of $\mathbb{Z}_2[x]$ modulo the one irreducible factor $x + 1$. For instance $[x][x^3] = [x^4] = [1]$, while $[x + 1][x^3 + x^2 + x + 1] = [(x + 1)^4] = 0$, and $[x^2 + 1] = [x + 1]^2$ is a nonzero element whose square is zero.
The modulus $x^4 + x^3 + 1$ does not factor: the previous unit proved it irreducible over $\mathbb{Z}_2$, as the reduction modulo 2 of $x^4 + 3x^3 + 1$. Being irreducible, it has no nonconstant factor in common with a nonzero polynomial of degree less than 4, so the gcd theorem predicts that all 15 nonzero classes are units, without computing a single inverse.
The script confirms both predictions by brute force: 15 units on the left, and on the right 8 units, exactly the classes with an odd number of terms. Nothing like the nonzero square $[x^2 + 1]^2 = 0$ can happen in a field, and the next article proves that the left-hand ring is one.
Parallels with the integers
| $\mathbb{Z}_n = \mathbb{Z}/(n)$ | $F[x]/(p(x))$ |
|---|---|
| $n$ elements, represented by $0, \ldots, n - 1$ | represented by the polynomials of degree less than $d$ (and 0); $q^d$ elements over a field with $q$ elements |
| operations well defined by the theorem on congruences | the same theorem for polynomials |
| every element is a sum $[1] + \cdots + [1]$ | every element is a polynomial expression in $\alpha = [x]$ over $F$, and $p(\alpha) = 0$ |
| $[a]$ a unit if and only if $\gcd(a, n) = 1$ | $[h]$ a unit if and only if $\gcd(h, p) = 1$ |
| zero divisors from a factorization of $n$ | zero divisors from a factorization of $p(x)$ |
Where it leads
- When $p(x)$ is irreducible, every nonzero class is relatively prime to it and so a unit: $F[x]/(p(x))$ is a field, it contains a root of $p(x)$, and over $\mathbb{Z}_p$ it is a finite field with $p^d$ elements (Building Fields from Irreducible Polynomials).
- The map $\varphi$ above is a ring homomorphism, and the identifications with $\mathbb{C}$ and $\mathbb{Q}(\sqrt 2)$ are the leading examples of the first isomorphism theorem.
- The set of multiples of $p(x)$ is an ideal, and $F[x]/(p(x))$ is a quotient ring; Congruence Modulo an Ideal and Quotient Rings carries the construction to any commutative ring and any ideal.
What we have learned
The classes modulo $p(x)$ form a commutative ring with identity, $F[x]/(p(x))$, built exactly as $\mathbb{Z}_n$ was. It contains a copy of $F$ and the class $\alpha = [x]$, with $p(\alpha) = 0$. Evaluation at a root identifies $\mathbb{R}[x]/(x^2 + 1)$ with $\mathbb{C}$, which is why the opening product came out complex. A class is a unit exactly when it is relatively prime to the modulus, and every other nonzero class is a zero divisor.
The sixteen-element ring with the irreducible modulus had no zero divisors at all. Is that true for every irreducible modulus? The next article, Building Fields from Irreducible Polynomials, proves that the ring is then a field, and uses it to give polynomials roots they did not have.
References
- Hungerford, T. W. (2014). Abstract Algebra: An Introduction (3rd ed.), §5.2 and the end of §5.3. Brooks/Cole, Cengage Learning. The course text: the operations on classes are well defined (Theorem 5.6); $F[x]/(p(x))$ is a commutative ring with identity containing a subring isomorphic to $F$ (Theorem 5.7), identified with $F$ (Theorem 5.8); classes relatively prime to the modulus are units (Theorem 5.9); $\mathbb{R}[x]/(x^2 + 1)$ and $\mathbb{C}$. record · 44 pages
- Judson, T. W. (2026). Abstract Algebra: Theory and Applications (annual edition, August 2026). Proposition 21.1.12 ($F(\alpha) \cong F[x]/\langle p(x) \rangle$ for the minimal polynomial $p(x)$ of $\alpha$, by the evaluation homomorphism and the first isomorphism theorem) and Example 21.1.14 ($\mathbb{R}[x]/\langle x^2 + 1 \rangle$ identified with $\mathbb{C}$). Free online at judsonbooks.org. record · 44 pages