Take a polynomial with integer coefficients, reduce its coefficients modulo 2, and divide the result by $x^2 + x + 1$ in $\mathbb{Z}_2[x]$. The remainder names a class in the four-element field $K = \mathbb{Z}_2[x]/(x^2 + x + 1)$ of Building Fields from Irreducible Polynomials. The rule $\varphi(f) = [\bar f]$ defines a homomorphism $\varphi: \mathbb{Z}[x] \to K$, as checked below. Since $x^3 = (x^2 + x + 1)(x + 1) + 1$ over $\mathbb{Z}_2$, it sends $x^3 + x$ to $[x + 1]$, and it sends $x + 3$ to $[x + 1]$ as well. The two inputs differ by $x^3 - 3 = (x^2 + x + 1)(x - 1) - 2$, which $\varphi$ sends to 0, and which lies in the ideal $J = (2, x^2 + x + 1)$ of the first article. Is that an accident? And if inputs with equal values always differ by an element of $J$, is the quotient ring $\mathbb{Z}[x]/J$ simply $K$ under another name?
It is no accident. Since $\varphi$ respects subtraction, $\varphi(f) = \varphi(g)$ exactly when $\varphi(f - g) = 0$, and the elements sent to 0 are, by the description of $J$ in the first article, exactly the elements of $J$. So the values of $\varphi$ correspond one for one with the cosets of $J$, the correspondence respects sums and products, and $\mathbb{Z}[x]/J \cong K$.
The argument used only that $\varphi$ is a homomorphism. The set of elements a homomorphism sends to 0 is its kernel. It is always an ideal, and the quotient by it is always isomorphic to the image. That is the first isomorphism theorem: to identify $R/I$, find a homomorphism out of $R$ whose kernel is $I$ and whose image is known.
First we show that kernels are ideals and decide when a map is one-to-one. Then we prove the theorem and apply it to quotients met earlier, to the Gaussian integers modulo $n$ and to a prime ideal of $\mathbb{Z}[x]$ that is not maximal. Finally it gives the Chinese remainder theorem its general form.
This is the third article of Ideals and Quotient Rings. It covers the second half of §6.2 of the course text (Hungerford, 2014), with Judson (2026, §16.3) as a free companion, and the Chinese remainder theorem for rings from the course text's optional §14.3. It uses homomorphisms and their first properties, ideals and quotient rings. Throughout, $f: R \to S$ is a homomorphism of rings.
Kernels
Definition. The kernel of $f$ is
$$\ker f = \{\, r \in R : f(r) = 0_S \,\}.$$
It always contains $0_R$, because $f(0_R) = 0_S$ (first properties).
Theorem. If $L$ is an ideal of $S$, then its preimage $f^{-1}(L) = \{\, r \in R : f(r) \in L \,\}$ is an ideal of $R$. In particular, $\ker f$, the preimage of the zero ideal $\{0_S\}$, is an ideal of $R$.
The idea: $f$ carries the two properties of $L$ back to $R$. Differences go to differences, and products go to products, which $L$ absorbs.
Proof. Apply the ideal test to $f^{-1}(L)$. It contains $0_R$, since $f(0_R) = 0_S \in L$. If $f(a)$ and $f(b)$ lie in $L$, then so does $f(a - b) = f(a) - f(b)$, because $L$ is closed under subtraction. If $f(a) \in L$ and $r \in R$, then $f(ra) = f(r) f(a)$ and $f(ar) = f(a) f(r)$ lie in $L$, because $L$ absorbs products. The zero ideal $\{0_S\}$ is an ideal of $S$, and its preimage is $\ker f$. $\blacksquare$
For the kernel this is the course text's Theorem 6.10 and Judson's Proposition 16.3.8. The running ideal is a preimage of this kind: $J$ is the set of polynomials whose reduction modulo 2 lies in the ideal $(x^2 + x + 1)$ of $\mathbb{Z}_2[x]$. The image $f(R)$, by contrast, is a subring of $S$ but usually not an ideal: the inclusion of $\mathbb{Z}$ in $\mathbb{Q}$ has image $\mathbb{Z}$.
Theorem. For $a, b \in R$, $f(a) = f(b)$ if and only if $a - b \in \ker f$, that is, if and only if $a + \ker f = b + \ker f$. In particular, $f$ is one-to-one if and only if $\ker f = \{0_R\}$.
The idea is the opening computation: equal values means the difference has value 0.
Proof. $f(a) = f(b)$ is equivalent to $f(a) - f(b) = 0$, that is, $f(a - b) = 0$. If $\ker f = \{0\}$, then $f(a) = f(b)$ forces $a - b = 0$, so $f$ is one-to-one. Conversely, if $f$ is one-to-one and $f(a) = 0 = f(0)$, then $a = 0$. $\blacksquare$
This is the course text's Theorem 6.11. It reduces the test for one-to-one maps to a single equation: instead of comparing all pairs, check which elements go to 0. The level sets of $f$, the sets of elements with a common value, are exactly the cosets of the kernel. For $\varphi$, the level set of $[x + 1]$ contains $x + 3$, $x^3 + x$ and every other element of the coset $(x + 1) + J$.
Examples
| Homomorphism | Kernel |
|---|---|
| $\mathbb{Z} \to \mathbb{Z}_n$, $a \mapsto [a]$ | $(n)$ |
| $\mathbb{Z}_n \to \mathbb{Z}_d$ for $d \mid n$, $[a]_n \mapsto [a]_d$ | $([d]_n)$ |
| $\mathbb{Z}[x] \to K$, $f \mapsto [\bar f]$ | $J = (2, x^2 + x + 1)$ |
| $\mathbb{Q}[x] \to \mathbb{R}$, $f \mapsto f(\sqrt 2)$ | $(x^2 - 2)$ |
| $R \to R/I$, the natural homomorphism | $I$ |
| $\mathbb{Z} \to R$, $n \mapsto n1_R$ | $(c)$, $c$ the characteristic of $R$ |
| the Frobenius map of $\mathbb{Z}_3[i]$ | $\{0\}$ |
The first is the reduction map. The second is the change of modulus, and its kernel consists of the classes of multiples of $d$, which the script checks for every divisor of 45. The kernels of the next two were found in Ideals and Principal Ideal Domains: the kernel of $\varphi$ is $J$ by its definition there, and the kernel of evaluation at $\sqrt 2$ is $(x^2 - 2)$. The natural homomorphism shows that every ideal is a kernel. The kernel of $n \mapsto n1_R$ is the set of $n$ with $n1_R = 0$, which the article on homomorphisms showed to be the multiples of the characteristic $c$; when $c = 0$ the kernel is $(0) = \{0\}$. The Frobenius map of $\mathbb{Z}_3[i]$ is the bijection $a + bi \mapsto a - bi$, and its kernel is $\{0\}$.
Check your understanding. The map $\varphi$ is the reduction $f \mapsto \bar f$ from $\mathbb{Z}[x]$ to $\mathbb{Z}_2[x]$ followed by the natural homomorphism onto $K$. Predict the kernel of the first step alone, and say how it sits inside $J$.
It is the principal ideal $(2)$, the polynomials whose coefficients are all even, since $\bar f = 0$ means every coefficient is divisible by 2. Anything the first step sends to 0, $\varphi$ also sends to 0, so $(2) \subseteq J$. The inclusion is strict: $x^2 + x + 1$ lies in $J$ but has odd coefficients. Composing with a second map can only enlarge the kernel.
Homomorphisms out of a field
A field has only two ideals, so a homomorphism out of a field has only two possible kernels.
Proposition. Let $F$ be a field and $f: F \to S$ a homomorphism. Then either $f$ sends every element to 0, or $f$ is one-to-one.
Proof. $\ker f$ is an ideal of $F$, so it is $\{0\}$ or $F$ (ideals containing a unit). In the first case $f$ is one-to-one; in the second it is the zero map. $\blacksquare$
The proposition explains a computation of the previous unit, where the Frobenius map $\Phi(a) = a^3$ of the nine-element field $\mathbb{Z}_3[i]$ turned out, element by element, to be the bijection $a + bi \mapsto a - bi$. In any finite field of characteristic $p$, the Frobenius map $\Phi(a) = a^p$ is a homomorphism (by the proposition proved there) and is not the zero map, since $\Phi(1) = 1$. So it is one-to-one, and a one-to-one function from a finite set to itself is onto. Every finite field therefore has the Frobenius map as an automorphism, with no computation. The script confirms the instance for the 49-element field $\mathbb{Z}_7[i]$ met below: $(a + bi)^7 = a - bi$ for all 49 elements, for example $(2 + 3i)^7 = 2 - 3i = 2 + 4i$.
The first isomorphism theorem
Here is the opening argument for every homomorphism.
Theorem (first isomorphism theorem). Let $f: R \to S$ be a homomorphism of rings with kernel $K$. Then the rule
$$\bar f(a + K) = f(a)$$
defines an isomorphism $\bar f$ from $R/K$ onto the image $f(R)$, and $f = \bar f \circ \pi$, where $\pi: R \to R/K$ is the natural homomorphism. In particular, if $f$ is onto, then $R/K \cong S$.
The idea: $f$ forgets which element of a coset of $K$ it was given, and nothing more, so cosets and values match one for one.
Proof. Well defined and one-to-one. By the previous theorem, $a + K = b + K$ if and only if $f(a) = f(b)$. Read from left to right, this says that the value $\bar f(a + K)$ does not depend on the representative $a$, so $\bar f$ is a function. Read from right to left, it says that $\bar f$ is one-to-one.
A homomorphism. For cosets $a + K$ and $b + K$,
$$\bar f\big((a + K) + (b + K)\big) = \bar f\big((a + b) + K\big) = f(a + b) = f(a) + f(b) = \bar f(a + K) + \bar f(b + K),$$
and the same computation with products in place of sums shows $\bar f\big((a + K)(b + K)\big) = \bar f(a + K)\,\bar f(b + K)$.
Onto $f(R)$. Every element of $f(R)$ is $f(a) = \bar f(a + K)$ for some $a$.
So $\bar f$ is a bijective homomorphism from $R/K$ to $f(R)$, an isomorphism. Finally $\bar f(\pi(a)) = \bar f(a + K) = f(a)$ for every $a$. $\blacksquare$
The course text states the theorem for onto homomorphisms (Theorem 6.13). Judson (2026, Theorem 16.3.12) states it with the image, as here, and adds that $\bar f$ is the only isomorphism with $f = \bar f \circ \pi$. That holds because $\pi$ is onto, so the values of $\bar f$ are forced. The factorization $f = \bar f \circ \pi$ describes every homomorphism as two steps: first collapse each coset of the kernel to a point, which is the only information $f$ loses, and then rename the result without loss.
Check your understanding. Evaluation at $\sqrt 2$, from $\mathbb{Q}[x]$ to $\mathbb{R}$, has kernel $(x^2 - 2)$. Does the theorem say that $\mathbb{Q}[x]/(x^2 - 2) \cong \mathbb{R}$?
No. The map is not onto, so the theorem identifies the quotient with the image $\mathbb{Q}[\sqrt 2]$, the numbers $a + b\sqrt 2$ with $a, b$ rational, not with $\mathbb{R}$.
How to use it
To show that $R/I \cong T$ for a given ideal $I$ and a known ring $T$:
- write down a homomorphism $f: R \to T$;
- show that $f$ is onto;
- show that $\ker f = I$. Usually one inclusion, $I \subseteq \ker f$, is a direct check on the generators of $I$, and the other needs an argument such as a division.
The theorem then supplies the isomorphism $a + I \mapsto f(a)$, with no need to check by hand that it is well defined or respects the operations. The examples below follow this recipe.
Quotients recognized
The running example: $\mathbb{Z}[x]$ modulo $J$
Back to the opening map.
Proposition. The map $\varphi(f) = [\bar f]$ is a homomorphism from $\mathbb{Z}[x]$ onto the four-element field $K = \mathbb{Z}_2[x]/(x^2 + x + 1)$ with kernel $J = (2, x^2 + x + 1)$. So
$$\mathbb{Z}[x]/J \cong K, \qquad f + J \mapsto [\bar f],$$
and $J$ is a maximal ideal of $\mathbb{Z}[x]$.
The idea: $\varphi$ is two known homomorphisms in a row, and its kernel is $J$ by the way $J$ was defined.
Proof. A homomorphism. Reduction modulo 2 is a homomorphism from $\mathbb{Z}[x]$ to $\mathbb{Z}_2[x]$ (reduction modulo a prime), and the natural map from $\mathbb{Z}_2[x]$ onto $K$ is one (the natural homomorphism). Their composite is $\varphi$, and a composite of homomorphisms is a homomorphism (Ring Homomorphisms and Isomorphisms). Onto. The four elements of $K$ are the classes of $0, 1, x$ and $x + 1$, and these integer polynomials are their own reductions. The kernel. $\varphi(f) = 0$ exactly when $x^2 + x + 1$ divides $\bar f$ in $\mathbb{Z}_2[x]$, which is the definition of $J$; the first article showed that $J = (2, x^2 + x + 1)$. The first isomorphism theorem gives the isomorphism. Finally, $x^2 + x + 1$ has no root in $\mathbb{Z}_2$, so being of degree 2 it is irreducible there, and $K$ is a field (Building Fields from Irreducible Polynomials). A quotient that is a field comes from a maximal ideal. $\blacksquare$
So the informal claim that opened Congruence Modulo an Ideal and Quotient Rings is now proved: each coset of $J$ contains exactly one of $0, 1, x, 1 + x$, since $\varphi$ sends these four to the four different elements of $K$. In the opening example, $x + 3$ and $x^3 + x$ lie in the coset of $1 + x$. The ideal $J$ is maximal, generated by two elements and, as the first article showed, by no single one.
Check your understanding. Does $x^4 + 3x^2 + 5$ lie in $J$? Reduce first.
Yes. Modulo 2 it becomes $x^4 + x^2 + 1$, which is $(x^2 + x + 1)^2$ in $\mathbb{Z}_2[x]$, since the cross terms $2x^3$, $2x^2$ and $2x$ vanish there. Over the integers, the script finds $x^4 + 3x^2 + 5 = (x^2 + x + 1)^2 + 2(2 - x - x^3)$, an explicit combination of the two generators.
Evaluation at a root
Congruence-class arithmetic identified $\mathbb{R}[x]/(x^2 + 1)$ with $\mathbb{C}$ and $\mathbb{Q}[x]/(x^2 - 2)$ with $\mathbb{Q}(\sqrt 2)$ by checking sums and products of classes. The theorem does both at once and explains what they have in common.
Proposition. Let $F$ be a subfield of a field $E$ and $c \in E$, and let $\varepsilon: F[x] \to E$ be evaluation at $c$, $\varepsilon(f) = f(c)$. Write $F[c] = \{f(c) : f \in F[x]\}$ for its image. If some nonzero polynomial over $F$ has $c$ as a root, then $\ker \varepsilon = (p(x))$ for an irreducible $p(x)$, and
$$F[x]/(p(x)) \cong F[c],$$
so $F[c]$ is a field.
The idea: the kernel is principal, and the quotient sits inside a field, so its kernel is prime and hence maximal.
Proof. Evaluation at $c$ is a homomorphism (evaluation), so its kernel is an ideal of $F[x]$, hence principal: $\ker \varepsilon = (p(x))$ (principal ideal domains). It is nonzero by hypothesis, so $p \neq 0$. By the first isomorphism theorem, $F[x]/(p) \cong F[c]$. The ring $F[c]$ is a subring of the field $E$ containing 1, so it is an integral domain, and therefore $(p)$ is a nonzero prime ideal (prime ideals). In the principal ideal domain $F[x]$ a nonzero prime ideal is maximal. So $F[x]/(p)$, and with it $F[c]$, is a field, and $p(x)$ is irreducible. $\blacksquare$
The generator $p(x)$, taken monic, is the polynomial of least degree with $c$ as a root, called the minimal polynomial of $c$ over $F$. Two cases:
- $\mathbb{Q}$ and $\sqrt 2$. The previous article found the kernel of evaluation at $\sqrt 2$ to be $(x^2 - 2)$, so $\mathbb{Q}[x]/(x^2 - 2) \cong \mathbb{Q}[\sqrt 2] = \{a + b\sqrt 2 : a, b \in \mathbb{Q}\}$, and this ring is a field.
- $\mathbb{R}$ and $i$. Evaluation $\mathbb{R}[x] \to \mathbb{C}$ at $i$ is onto, since $a + bx \mapsto a + bi$. Its kernel contains $x^2 + 1$ and no nonzero polynomial of degree less than 2, since $a + bi = 0$ forces $a = b = 0$. So the kernel is $(x^2 + 1)$, and $\mathbb{R}[x]/(x^2 + 1) \cong \mathbb{C}$.
This is the converse direction of Kronecker's theorem. Kronecker started from an irreducible $p(x)$ and built a field $F[x]/(p(x))$ in which $p$ has a root. The proposition starts from a root $c$ in a known field and shows that the field it generates is of Kronecker's kind.
Changing the modulus
Proposition. If $d$ is a positive divisor of $n$, then $\mathbb{Z}_n/([d]) \cong \mathbb{Z}_d$.
The idea: reducing modulo $d$ forgets exactly the multiples of $d$.
Proof. The change of modulus $[a]_n \mapsto [a]_d$ is a homomorphism from $\mathbb{Z}_n$ onto $\mathbb{Z}_d$. Its kernel consists of the $[a]_n$ with $d \mid a$, which is the ideal $([d]_n)$ of $\mathbb{Z}_n$. $\blacksquare$
The cosets of $([d])$ described in Congruence Modulo an Ideal and Quotient Rings, the sets of elements with the same remainder on division by $d$, are exactly the level sets of this map. For the running ring $\mathbb{Z}_{45}$ the proposition confirms the quotients $\mathbb{Z}_d$ drawn beside the six ideals in the figure of the first article, and with them the list of prime and maximal ideals: $\mathbb{Z}_{45}/(d) \cong \mathbb{Z}_d$ is a field exactly when $d$ is 3 or 5.
The Gaussian integers modulo $n$
What do ordinary integers $n$ give as moduli in $\mathbb{Z}[i]$? The rings $\mathbb{Z}_n[i]$, pairs $a + bi$ with coordinates in $\mathbb{Z}_n$ and the rule $i^2 = -1$, were introduced as a family of finite rings. They are quotients of the Gaussian integers.
Proposition. For $n \ge 2$, the map $\rho(a + bi) = [a] + [b]i$ is a homomorphism from $\mathbb{Z}[i]$ onto $\mathbb{Z}_n[i]$ with kernel $(n)$. So $\mathbb{Z}[i]/(n) \cong \mathbb{Z}_n[i]$.
The idea: reduce both coordinates modulo $n$.
Proof. The sum and product of $a + bi$ and $c + di$ are given by the formulas $(a + c) + (b + d)i$ and $(ac - bd) + (ad + bc)i$ in both rings, with the coordinates reduced modulo $n$ in $\mathbb{Z}_n[i]$. Reduction modulo $n$ respects sums and products of integers, so $\rho$ respects both operations. It is onto, since $[a] + [b]i$ is the image of $a + bi$. Its kernel consists of the $a + bi$ with $n \mid a$ and $n \mid b$. These are exactly the multiples of $n$ in $\mathbb{Z}[i]$: a multiple $n(c + di) = nc + ndi$ has both coordinates divisible by $n$, and $nc + ndi$ is the multiple $n(c + di)$. $\blacksquare$
With the theorem of Integral Domains and Fields that $\mathbb{Z}_p[i]$ is a field exactly when $p \equiv 3 \pmod 4$, for $p$ an odd prime, this decides which primes of $\mathbb{Z}$ generate maximal ideals of $\mathbb{Z}[i]$.
Corollary. For an odd prime $p$, the ideal $(p)$ of $\mathbb{Z}[i]$ is maximal if and only if $p \equiv 3 \pmod 4$. For $p \equiv 1 \pmod 4$, and for $p = 2$, it is not even prime.
Proof. $\mathbb{Z}[i]/(p) \cong \mathbb{Z}_p[i]$ is a field exactly when $p \equiv 3 \pmod 4$. In the remaining cases $\mathbb{Z}_p[i]$ has zero divisors (for $p = 2$, $(1 + i)^2 = 2i = 0$), so it is not an integral domain and $(p)$ is not prime. $\blacksquare$
The script confirms the three cases it is given.
- For $p = 7$, all 48 nonzero elements of $\mathbb{Z}_7[i]$ are units, so $\mathbb{Z}[i]/(7)$ is a field with 49 elements; $-1$ is not a square modulo 7.
- For $p = 13$, $-1 \equiv 5^2 \equiv 8^2$, and the factorization $13 = (3 + 2i)(3 - 2i)$ shows directly that $(13)$ is not prime: the product lies in $(13)$, but neither factor is a multiple of 13 in $\mathbb{Z}[i]$, since $\tfrac{3}{13}$ is not an integer. The script counts 144 units and 24 zero divisors in $\mathbb{Z}_{13}[i]$.
- For $p = 2$, $(1 + i)^2 = 2i$ lies in $(2)$ while $1 + i$ does not; in $\mathbb{Z}_2[i]$, with four elements, $1 + i$ is a nonzero element whose square is 0.
Check your understanding. $\mathbb{Z}[i]/(2) \cong \mathbb{Z}_2[i]$ has four elements, and so does $\mathbb{Z}_4$, which also has a nonzero element with square 0. Are the two rings isomorphic?
No. $\mathbb{Z}_2[i]$ has characteristic 2 and $\mathbb{Z}_4$ has characteristic 4, and an isomorphism preserves the characteristic.
A prime ideal that is not maximal
In a principal ideal domain the nonzero prime ideals are maximal (Congruence Modulo an Ideal and Quotient Rings). The polynomial ring $\mathbb{Z}[x]$ is not a principal ideal domain, and the Gaussian integers supply a prime ideal of it that is not maximal.
Proposition. Evaluation at $i$ is a homomorphism from $\mathbb{Z}[x]$ onto $\mathbb{Z}[i]$ with kernel $(x^2 + 1)$. So $\mathbb{Z}[x]/(x^2 + 1) \cong \mathbb{Z}[i]$, and $(x^2 + 1)$ is a prime ideal of $\mathbb{Z}[x]$ that is not maximal.
The idea: the remainder $c + dx$ on division by $x^2 + 1$ has value $c + di$ at $i$, so value 0 means remainder 0.
Proof. $\mathbb{Z}$ is a subring of the commutative ring $\mathbb{Z}[i]$, so evaluation at $i$ is a homomorphism (evaluation). It is onto, since $a + bx \mapsto a + bi$. Every multiple of $x^2 + 1$ goes to 0, since $i^2 + 1 = 0$. Conversely, let $f(i) = 0$. Dividing $f$ by the monic polynomial $x^2 + 1$ in $\mathbb{Z}[x]$ gives $f = (x^2 + 1)q + (c + dx)$ with integers $c$ and $d$, and then $0 = f(i) = c + di$, so $c = d = 0$ and $f \in (x^2 + 1)$. The theorem gives the isomorphism. The Gaussian integers are an integral domain but not a field (2 has no inverse, since $2(a + bi) = 1$ would need $2a = 1$), so the ideal is prime and not maximal. $\blacksquare$
For example, $f = x^4 + 3x^3 + x + 5$ has $f(i) = 1 - 3i + i + 5 = 6 - 2i$, and the script's division gives $f = (x^2 + 1)(x^2 + 3x - 1) + (6 - 2x)$: the remainder, read with $x = i$, is the value.
Check your understanding. Without dividing, decide whether $x^3 + x^2 + x + 1$ lies in $(x^2 + 1)$.
It does, since its value at $i$ is $-i - 1 + i + 1 = 0$. The script's division agrees: it is $(x^2 + 1)(x + 1)$.
The ideal $(x^2 + 1)$ lies inside a maximal ideal. Compose evaluation at $i$ with the map $\rho$ of the previous section for $n = 7$: $f \mapsto f(i)$ reduced modulo 7 is a homomorphism from $\mathbb{Z}[x]$ onto $\mathbb{Z}_7[i]$. Its kernel is $(7, x^2 + 1)$. Indeed, a combination $7g + (x^2 + 1)h$ goes to $7g(i)$, whose coordinates are multiples of 7. Conversely, suppose $f(i) = c + di$ has $7 \mid c$ and $7 \mid d$. Dividing as before, $f = (x^2 + 1)q + (c + dx)$, and $c + dx = 7(c/7 + (d/7)x)$, so $f \in (7, x^2 + 1)$. For instance $x^4 + 7x + 6 = (x^2 + 1)(x^2 - 1) + 7(x + 1)$, with value $7 + 7i$ at $i$. Since $\mathbb{Z}_7[i]$ is a field,
$$(x^2 + 1) \subsetneq (7, \, x^2 + 1) \subsetneq \mathbb{Z}[x],$$
with quotients $\mathbb{Z}[i]$, an integral domain, and $\mathbb{Z}_7[i]$, a field with 49 elements.
The Chinese remainder theorem for rings
The Chinese remainder theorem was proved twice, for integers and for polynomials, with the same three ingredients:
- Bezout's identity $ms + nt = 1$;
- the idempotents $e_1 = nt$ and $e_2 = ms$;
- a counting or divisibility argument for one-to-one.
The first isomorphism theorem turns the two proofs into one, valid in any commutative ring, with the kernel in place of the third ingredient.
Theorem. Let $R$ be a commutative ring with identity and $A$, $B$ ideals of $R$ for which there are $a \in A$ and $b \in B$ with $a + b = 1$. Then
$$R/(A \cap B) \cong R/A \times R/B, \qquad r + (A \cap B) \mapsto (r + A, \; r + B).$$
The idea: send $r$ to its pair of cosets; the split $1 = a + b$ builds an element with any prescribed pair.
Proof. Let $h(r) = (r + A, r + B)$. Each coordinate is a natural homomorphism, so $h$ is a homomorphism into the direct product (by the lemma on maps into a product in The Chinese Remainder Theorem). Its kernel consists of the $r$ with $r + A = A$ and $r + B = B$, that is, $r \in A$ and $r \in B$. So $\ker h = A \cap B$; in particular $A \cap B$ is an ideal, being a kernel. For onto, the element $b = 1 - a$ satisfies $b + A = 1 + A$, since $1 - b = a \in A$, and $b + B = 0 + B$, since $b \in B$. Likewise $a + A = 0 + A$ and $a + B = 1 + B$. Given cosets $u + A$ and $v + B$, the element $r = ub + va$ has
$$h(r) = (u + A, \; 0 + B) + (0 + A, \; v + B) = (u + A, \; v + B),$$
using that $h$ respects sums and products. The first isomorphism theorem gives the isomorphism. $\blacksquare$
The elements $b$ and $a$ play the roles of the idempotents $e_1$ and $e_2$, and the hypothesis $a + b = 1$ replaces Bezout's identity. For $R = \mathbb{Z}$, $A = (m)$ and $B = (n)$ with $ms + nt = 1$, take $a = ms$ and $b = nt$. The intersection $(m) \cap (n)$ consists of the common multiples of $m$ and $n$, which are the multiples of $mn$ when $\gcd(m, n) = 1$ (lemma on relatively prime moduli). The theorem becomes $\mathbb{Z}_{mn} \cong \mathbb{Z}_m \times \mathbb{Z}_n$. The course text treats this version in §14.3, and Judson (2026, §16.7, Exercise 39) poses it.
An example in $\mathbb{Z}[x]$. Take $A = (x^2 + 1)$ and $B = (x^2 + x + 1)$. The two quadratics differ by $x$, and $(x^2 + 1) - x \cdot x = 1$. Substituting $x = (x^2 + x + 1) - (x^2 + 1)$ gives
$$(1 + x)(x^2 + 1) - x(x^2 + x + 1) = 1.$$
So $a = (1 + x)(x^2 + 1) = x^3 + x^2 + x + 1$ lies in $A$, $b = -x(x^2 + x + 1) = -x^3 - x^2 - x$ lies in $B$, and $a + b = 1$. The intersection $A \cap B$ is the principal ideal generated by the product
$$p = (x^2 + 1)(x^2 + x + 1) = x^4 + x^3 + 2x^2 + x + 1.$$
Certainly $p$ lies in both ideals. Conversely, let $f$ lie in both, and write $f = fa + fb$. In $fa$, $f$ is a multiple of $x^2 + x + 1$ and $a$ is a multiple of $x^2 + 1$, so $fa$ is a multiple of $p$; likewise $fb$. So $f \in (p)$. The theorem, together with the identification of $\mathbb{Z}[x]/(x^2 + 1)$ with $\mathbb{Z}[i]$ in the previous section, gives
$$\mathbb{Z}[x]/(p) \cong \mathbb{Z}[i] \times \mathbb{Z}[x]/(x^2 + x + 1), \qquad f + (p) \mapsto \big(f(i), \; f + (x^2 + x + 1)\big).$$
Each class modulo $p$ is named by its remainder on division by the monic $p$, of degree less than 4. The pair records the value at $i$ and the remainder on division by $x^2 + x + 1$. For example $x^5$ leaves the remainder $-x^3 + x^2 + 1$ modulo $p$. Both have the value $i$ at $i$, and both leave the remainder $-x - 1$ modulo $x^2 + x + 1$, as the script finds. The class of $a$ corresponds to $(0, 1)$ and the class of $b$ to $(1, 0)$. They are idempotents, $a^2 \equiv a$, with product $ab \equiv 0$ and sum 1, the counterparts of 10 and 36 in $\mathbb{Z}_{45}$. The script checks that products of remainders with coefficients from $-1$ to 1 match the coordinatewise products in all 6,561 cases.
Check your understanding. Which polynomial of degree less than 4 corresponds to the pair $(i, 0)$? Use the proof's recipe $r = ub + va$.
Take $u = x$, which has value $i$ at $i$, and $v = 0$. Then $r = xb = -x^4 - x^3 - x^2$, and adding $p$ leaves the remainder $x^2 + x + 1$. That is easy to confirm: $x^2 + x + 1$ is 0 modulo itself, and its value at $i$ is $-1 + i + 1 = i$.
Every image is a quotient
The theorem says that the homomorphic images of a ring $R$, up to isomorphism, are exactly its quotients $R/I$. Every image is a quotient by the kernel, and every quotient is the image of the natural homomorphism. Knowing the ideals of $R$ therefore means knowing every ring that $R$ can be mapped onto.
- Images of $\mathbb{Z}$. The ideals of $\mathbb{Z}$ are the $(n)$, so the images of $\mathbb{Z}$ are $\mathbb{Z}$ and the rings $\mathbb{Z}_n$. Applied to $n \mapsto n1_R$ in a ring $R$ with identity, the theorem says that the multiples of $1_R$ form a copy of $\mathbb{Z}$ (characteristic 0) or of $\mathbb{Z}_c$ (characteristic $c$). That is the dichotomy that Ring Homomorphisms and Isomorphisms found by a direct argument. The script records the cases $\mathbb{Z}_{45}$ (kernel $(45)$), $\mathbb{Z}_7[i]$ (kernel $(7)$) and $\mathbb{Z} \times \mathbb{Z}$ (kernel $(0)$).
- Images of $\mathbb{Z}_{45}$. The six ideals give the six images $\mathbb{Z}_d$, $d \mid 45$, as in the figure of the first article.
Check your understanding. Which copy of $\mathbb{Z}$ or $\mathbb{Z}_c$ do the multiples of 1 form in the field $\mathbb{Z}[x]/(5, x^2 - 3)$ of Congruence Modulo an Ideal and Quotient Rings?
A copy of $\mathbb{Z}_5$: 5 lies in the ideal, while $1, 2, 3, 4$ are distinct representatives, so the kernel of $n \mapsto n \cdot 1$ is $(5)$.
There are further isomorphism theorems, describing the quotient of a sum $(I + J)/J$ and quotients of quotients (Judson, 2026, Theorems 16.3.13 and 16.3.14). They follow from the first, and the course did not need them.
The same theorem for groups
Groups have homomorphisms, kernels and quotients too, and the theorem holds there with the same proof. The kernel of a group homomorphism is a normal subgroup, the subgroups one can work modulo. The first isomorphism theorem for groups identifies the quotient by the kernel with the image. Judson proves the group version first and derives the ring version from it, by applying the group theorem to the additive groups and then checking products. The proof here is self-contained, since groups come later in this course.
Where it leads
- Group homomorphisms and quotient groups repeat the theory of this unit with one operation instead of two.
- Fields of the form $F[x]/(p(x))$, now recognized as $F[c]$ for a root $c$, are the starting point of field theory and Galois theory, which study the automorphisms, such as the Frobenius map, of such fields.
- The Chinese remainder theorem for rings, with its applications, is developed further in the optional Chapter 14 of the course text.
What we have learned
A kernel is an ideal, and two elements have the same image exactly when they share a coset of it. The first isomorphism theorem makes $R/\ker f$ isomorphic to the image. So $\mathbb{Z}[x]/(2, x^2 + x + 1)$ is the field with four elements, $\mathbb{Z}[x]/(x^2 + 1)$ is $\mathbb{Z}[i]$ and $\mathbb{R}[x]/(x^2 + 1)$ is $\mathbb{C}$, and the Chinese remainder theorem holds in every commutative ring.
Cosets were defined with addition alone, and multiplication entered only through absorption. What becomes of cosets, quotients and this theorem in a structure with a single operation? The next unit, Groups, studies that structure from its first article, Groups: Definition and First Examples, and the course returns to cosets, quotients and this theorem in Cosets, Normal Subgroups and Quotient Groups.
References
- Hungerford, T. W. (2014). Abstract Algebra: An Introduction (3rd ed.). Brooks/Cole, Cengage Learning. The course text: §6.2, the kernel is an ideal (Theorem 6.10), a homomorphism is one-to-one exactly when its kernel is zero (Theorem 6.11), the natural homomorphism (Theorem 6.12), the first isomorphism theorem for onto homomorphisms (Theorem 6.13); §14.3, the Chinese remainder theorem for rings (cited from the section title only). record · 44 pages
- Judson, T. W. (2026). Abstract Algebra: Theory and Applications (annual edition, August 2026). Free online at judsonbooks.org. §16.3: kernels (Proposition 16.3.8), the first isomorphism theorem with the image and the uniqueness of the induced map (Theorem 16.3.12), the second and third isomorphism theorems (Theorems 16.3.13 and 16.3.14); §16.7, Exercise 39, the Chinese remainder theorem for rings. record · 44 pages