Ring Homomorphisms and Isomorphisms

Maps between rings that respect addition and multiplication. The definitions, with examples from reduction modulo n, evaluation of polynomials, a map from the Eisenstein integers onto Z_7 and the Frobenius map, and non-examples from determinants and transposes; the first properties of a homomorphism, proved in full, including what happens to the identity and to units; the homomorphisms out of Z and the characteristic of a ring; composites and inverses; the properties an isomorphism preserves, used to tell apart three rings with nine elements; and two rings with four elements and two with dual numbers shown to be the same ring under different names.

Here are two rings with four elements. The first is $\mathbb{Z}_2[x]/(x^2 + x + 1)$, the four-element field at the head of the table of small fields in Building Fields from Irreducible Polynomials. It is computed with the congruence-class arithmetic of Congruence-Class Arithmetic in Polynomial Rings, modulo an irreducible quadratic. Its elements are $0$, $1$, $\alpha$ and $1 + \alpha$, where $\alpha$ is the class of $x$ and $\alpha^2 = \alpha + 1$. The second is a set of four matrices with entries in $\mathbb{Z}_2$,

$$0, \qquad I = \begin{pmatrix} 1 & 0 \\ 0 & 1 \end{pmatrix}, \qquad A = \begin{pmatrix} 0 & 1 \\ 1 & 1 \end{pmatrix}, \qquad I + A = \begin{pmatrix} 1 & 1 \\ 1 & 0 \end{pmatrix},$$

with the matrix operations reduced modulo 2. Squaring $A$ gives $I + A$, so $A^2 = A + I$, the same rule that $\alpha$ obeys. The script checks that the four matrices are closed under sums and products. It also checks that the multiplication table of the matrices is the multiplication table of $\mathbb{Z}_2[x]/(x^2 + x + 1)$ with $\alpha$ renamed $A$:

$\times$ $0$ $1$ $\alpha$ $1 + \alpha$
$0$ $0$ $0$ $0$ $0$
$1$ $0$ $1$ $\alpha$ $1 + \alpha$
$\alpha$ $0$ $\alpha$ $1 + \alpha$ $1$
$1 + \alpha$ $0$ $1 + \alpha$ $1$ $\alpha$

The addition tables match in the same way. A polynomial class and a matrix are very different objects, yet the two rings cannot be told apart by computing in them. In what precise sense are they the same ring, and what does that sameness let us conclude?

The answer is an isomorphism: a renaming of elements that carries sums to sums and products to products. Anything proved about one ring from its tables, for example that it is a field, is then true of the other. Renamings are the special case of a broader idea, a homomorphism, which is a map between rings that respects both operations but may merge elements or miss some. Reducing an integer modulo $n$ is the first example, and the Chinese remainder theorem in the next article is the leading one.

First we state the definitions and try them on examples, including maps that respect one operation but not the other. Then we prove the first properties every homomorphism has, and find that it need not send 1 to 1. The homomorphisms out of $\mathbb{Z}$ lead to the characteristic of a ring. Finally we list what an isomorphism preserves, use the list to tell apart three rings with nine elements, and produce isomorphisms where they exist.

This article opens Homomorphisms and the Chinese Remainder Theorem. It covers §3.3 of the course text (Hungerford, 2014), with Judson (2026, §§16.2 and 16.3) as a free companion. It uses the rings of the Rings unit, the evaluation of polynomials, and the quotient rings $F[x]/(p(x))$ of Congruence-Class Arithmetic in Polynomial Rings.

Definitions

Definition. Let $R$ and $S$ be rings. A function $f: R \to S$ is a homomorphism if

$$f(a + b) = f(a) + f(b) \qquad \text{and} \qquad f(ab) = f(a)\,f(b)$$

for all $a$ and $b$ in $R$. A homomorphism that is a bijection (one-to-one and onto) is an isomorphism. The rings $R$ and $S$ are isomorphic, written $R \cong S$, if there is an isomorphism from $R$ to $S$. An isomorphism from a ring to itself is an automorphism.

On the left of each equation the operation is that of $R$; on the right it is that of $S$. The definition asks that it makes no difference whether two elements are combined first and the result carried over to $S$, or carried over first and combined in $S$.

The course text and Judson both use this definition, which does not ask that $f(1_R) = 1_S$. Many other books, particularly those in which every ring has an identity, add that requirement to the definition. The difference matters for a few statements below, which say when they need $f(1) = 1$.

Back to the two rings of four elements. The tables show that $a + b\alpha \mapsto aI + bA$, for $a$ and $b$ in $\mathbb{Z}_2$, is an isomorphism; the script checks all 16 sums and all 16 products. The modulus $x^2 + x + 1$ is irreducible over $\mathbb{Z}_2$, since it has no roots there, so $\mathbb{Z}_2[x]/(x^2 + x + 1)$ is a field. Then so is the ring of four matrices: a field whose elements are matrices.

Check your understanding. Without multiplying any matrices, predict $A^3$.

Read it off the table. In $\mathbb{Z}_2[x]/(x^2 + x + 1)$, $\alpha^3 = \alpha \cdot \alpha^2 = \alpha(1 + \alpha) = 1$. The isomorphism carries this equation to $A^3 = I$. Multiplying confirms it: $A^3 = A \cdot A^2 = A(I + A) = A + A^2 = A + A + I = I$, since $A + A = 0$ modulo 2. The script finds $A^3 = I$ as well. This is what "the same ring" buys: a fact found in one ring transfers to the other without new work.

Examples

Which familiar maps respect both operations? The first examples come from the earlier units, where several of them were already in use.

Reduction modulo $n$. The map $\mathbb{Z} \to \mathbb{Z}_n$, $a \mapsto [a]$, is a homomorphism. The equations $[a + b] = [a] + [b]$ and $[ab] = [a][b]$ are the definitions of the operations of $\mathbb{Z}_n$ (Judson, 2026, Example 16.3.1). The map is onto and, since $[0] = [n]$, not one-to-one.

Changing the modulus. If $d$ divides $n$, then $[a]_n \mapsto [a]_d$ is a well-defined function from $\mathbb{Z}_n$ to $\mathbb{Z}_d$. Indeed, if $[a]_n = [b]_n$, then $n \mid a - b$, so $d \mid a - b$ and $[a]_d = [b]_d$. It respects both operations, since $[a + b]_d = [a]_d + [b]_d$ and $[ab]_d = [a]_d [b]_d$. The script confirms the rule and both operations for every $n$ up to 30 and every divisor $d \ge 2$ of $n$, 81 pairs in all. When $d$ does not divide $n$, the rule is not a function at all: the article on $\mathbb{Z}_n$ shows that $[a]_6 \mapsto [a]_4$ gives two values for the class $[0]_6 = [6]_6$.

Evaluation. For a commutative ring $R$ and an element $c$ of $R$, evaluation $f(x) \mapsto f(c)$ is a homomorphism from $R[x]$ to $R$. This is the theorem on evaluation, whose proof needs commutativity. More generally, if $R$ is a subring of a commutative ring $E$ and $c \in E$, evaluation at $c$ is a homomorphism from $R[x]$ to $E$, by the same proof.

  • Evaluating rational polynomials at $\sqrt 2$ gives a homomorphism $\mathbb{Q}[x] \to \mathbb{R}$. Its values are the numbers $a + b\sqrt 2$ with $a, b$ rational, since $\sqrt 2^{\,2} = 2$ lets every higher power be rewritten.
  • When $c$ is a root of a modulus $p(x)$, evaluation passes to the classes modulo $p(x)$. This is how Congruence-Class Arithmetic in Polynomial Rings identified $\mathbb{R}[x]/(x^2 + 1)$ with $\mathbb{C}$ and $\mathbb{Q}[x]/(x^2 - 2)$ with $\mathbb{Q}(\sqrt 2)$.
  • Evaluating functions at a point is a homomorphism for the same reason: the operations on functions are defined pointwise (Judson, 2026, Example 16.3.2).

Inclusions, the identity map and the zero map. If $S$ is a subring of $R$, the inclusion $s \mapsto s$ is a one-to-one homomorphism from $S$ to $R$, because $S$ uses the operations of $R$. The identity map of any ring is an automorphism. The map that sends every element of $R$ to $0_S$ is a homomorphism, since $0 + 0 = 0$ and $0 \cdot 0 = 0$.

The Eisenstein integers onto $\mathbb{Z}_7$

The examples so far were familiar. Here is one that needs an idea. The Eisenstein integers $\mathbb{Z}[\omega] = \{\, a + b\omega : a, b \in \mathbb{Z} \,\}$ are built from a complex cube root of 1, $\omega = -\tfrac12 + \tfrac{\sqrt 3}{2} i$, which satisfies $\omega^2 = -1 - \omega$, that is, $\omega^2 + \omega + 1 = 0$. In $\mathbb{Z}_7$, $2^2 + 2 + 1 = 7 = 0$: the number 2 plays the part of $\omega$. This suggests sending $\omega$ to 2. Define $g: \mathbb{Z}[\omega] \to \mathbb{Z}_7$ by

$$g(a + b\omega) = [a + 2b].$$

Proposition. $g$ is a homomorphism from $\mathbb{Z}[\omega]$ onto $\mathbb{Z}_7$.

Here is the idea. Sums are no trouble, because $g$ is linear in $a$ and $b$. In a product, the only place where $\omega$ does anything special is $\omega^2 = -1 - \omega$, and $2^2 = 4 = -1 - 2$ in $\mathbb{Z}_7$ too, so the two computations differ by a multiple of 7.

Proof. Each Eisenstein integer is $a + b\omega$ for exactly one pair of integers $a, b$, since $\omega$ is not real, so $g$ is a function. For sums,

$$g\big((a + b\omega) + (c + d\omega)\big) = [(a + c) + 2(b + d)] = [a + 2b] + [c + 2d].$$

For products, $(a + b\omega)(c + d\omega) = (ac - bd) + (ad + bc - bd)\omega$ by the product formula of the subrings article, so

$$g\big((a + b\omega)(c + d\omega)\big) = [ac - bd + 2ad + 2bc - 2bd] = [ac + 2ad + 2bc - 3bd],$$

while

$$g(a + b\omega)\,g(c + d\omega) = [(a + 2b)(c + 2d)] = [ac + 2ad + 2bc + 4bd].$$

The two classes differ by $[7bd] = [0]$, so they are equal. The map is onto because $g(k) = [k]$ for each integer $k$. $\blacksquare$

For example, $(2 + 3\omega)(1 - \omega) = 5 + 4\omega$, and $g(5 + 4\omega) = [13] = [6]$, while $g(2 + 3\omega)\, g(1 - \omega) = [8][-1] = [1][6] = [6]$. The unit $\omega$ of $\mathbb{Z}[\omega]$ has inverse $\omega^2 = -1 - \omega$, since $\omega^3 = 1$. It goes to $[2]$, and its inverse goes to $g(-1 - \omega) = [-3] = [4]$, the inverse of 2, as $2 \cdot 4 = 8 = 1$ in $\mathbb{Z}_7$. Since $4^2 + 4 + 1 = 21 = 0$ in $\mathbb{Z}_7$ as well, sending $\omega$ to 4 instead gives a second homomorphism onto $\mathbb{Z}_7$, by the same proof.

Check your understanding. Replace 7 by another prime $n$ and 2 by a number $c$. For which $c$ is $a + b\omega \mapsto [a + cb]$ a homomorphism from $\mathbb{Z}[\omega]$ to $\mathbb{Z}_5$? To $\mathbb{Z}_{13}$?

Rerun the proof with $c$ in place of 2: the two classes for a product now differ by $[(c^2 + c + 1)bd]$. Taking $b = d = 1$ shows that the map respects products exactly when $c^2 + c + 1 = 0$ in $\mathbb{Z}_n$. In $\mathbb{Z}_5$, $c^2 + c + 1$ takes the values $1, 3, 2, 3, 1$ for $c = 0, \ldots, 4$, never 0, so no $c$ works. In $\mathbb{Z}_{13}$, $3^2 + 3 + 1 = 13 = 0$ and $9^2 + 9 + 1 = 91 = 7 \cdot 13 = 0$, so $c = 3$ and $c = 9$ both work. The script confirms both answers. The two roots add to $12 = -1$ in $\mathbb{Z}_{13}$, as 2 and 4 add to $6 = -1$ in $\mathbb{Z}_7$, mirroring $\omega + \omega^2 = -1$.

The Eisenstein integers a + bω near 0, drawn in the complex plane and each labeled with its image a + 2b in Z7. Notice that the points sent to 0 (orange) form a triangular grid spanned by 2 - ω and 1 + 3ω = ω(2 - ω), and that each other value fills a shifted copy of that grid. Moving one step right (adding 1) adds 1 to the label; moving by ω, up and to the left, adds 2.
The Eisenstein integers a + bω near 0, drawn in the complex plane and each labeled with its image a + 2b in Z_7. Notice that the points sent to 0 (orange) form a triangular grid spanned by 2 - ω and 1 + 3ω = ω(2 - ω), and that each other value fills a shifted copy of that grid. Moving one step right (adding 1) adds 1 to the label; moving by ω, up and to the left, adds 2.

The figure shows which Eisenstein integers $g$ sends to 0. They are exactly the multiples of $2 - \omega$ in $\mathbb{Z}[\omega]$.

  • A multiple $(2 - \omega)(c + d\omega) = (2c + d) + (3d - c)\omega$ goes to $[2c + d + 6d - 2c] = [7d] = [0]$.
  • Conversely, if $a + 2b = 7d$ for an integer $d$, put $c = 3d - b$. Then $2c + d = 7d - 2b = a$ and $3d - c = b$, so $a + b\omega = (2 - \omega)(c + d\omega)$.

The multiples of $2 - \omega$ include $\omega(2 - \omega) = 2\omega - \omega^2 = 1 + 3\omega$. Every multiple $(2 - \omega)(c + d\omega) = c(2 - \omega) + d(1 + 3\omega)$ is an integer combination of these two, which is the grid of the figure. The set of elements a homomorphism sends to 0 is its kernel. Kernels are taken up in the unit on Ideals and Quotient Rings, and the first isomorphism theorem turns the map $g$ into an isomorphism between $\mathbb{Z}_7$ and the Eisenstein integers modulo $2 - \omega$.

Maps that respect one operation but not the other

Both conditions in the definition are needed, and natural maps often satisfy only one.

Determinant. On $2 \times 2$ real matrices, $\det(XY) = \det X \det Y$, so the determinant respects multiplication. It does not respect addition: $\det(I + I) = \det(2I) = 4$, while $\det I + \det I = 2$.

Transpose. The transpose respects addition, $(X + Y)^{\mathsf T} = X^{\mathsf T} + Y^{\mathsf T}$, but reverses products: $(XY)^{\mathsf T} = Y^{\mathsf T} X^{\mathsf T}$. With $X = \begin{pmatrix} 1 & 1 \\ 0 & 1 \end{pmatrix}$ and $Y = \begin{pmatrix} 1 & 0 \\ 1 & 1 \end{pmatrix}$, the script finds $(XY)^{\mathsf T} = \begin{pmatrix} 2 & 1 \\ 1 & 1 \end{pmatrix}$ and $X^{\mathsf T} Y^{\mathsf T} = \begin{pmatrix} 1 & 1 \\ 1 & 2 \end{pmatrix}$. So the transpose is not a homomorphism from $M_2(\mathbb{R})$ to itself, although it is a bijection that preserves a great deal.

Adding a constant. On $\mathbb{Z}$, $h(a) = a + 1$ fails both conditions; the quickest check is $h(0 + 0) = 1$ but $h(0) + h(0) = 2$. Every homomorphism sends 0 to 0, as proved below, so a map that moves 0 is ruled out at once.

The Frobenius map

Squaring is a natural candidate of the same kind. It respects multiplication in any commutative ring, since $(ab)^2 = a^2 b^2$, but in general not addition, since $(a + b)^2 = a^2 + 2ab + b^2$. In a commutative ring in which $2a = 0$ for every $a$, such as $\mathbb{Z}_2 \times \mathbb{Z}_2 \times \mathbb{Z}_2$, the middle term vanishes and squaring is a homomorphism. The same happens with $p$-th powers for any prime $p$.

Proposition. Let $p$ be a prime and $R$ a commutative ring in which $pa = 0$ for every $a$. Then $\Phi(a) = a^p$ is a homomorphism from $R$ to $R$.

Here is the idea. Expand $(a + b)^p$ by the binomial theorem. Every term except the first and last carries a binomial coefficient divisible by $p$, and $p$ times anything is 0 in $R$.

Proof. Products: $(ab)^p = a^p b^p$ by commutativity. Sums: by the binomial theorem, which holds in every commutative ring,

$$(a + b)^p = a^p + \sum_{k=1}^{p-1} \binom{p}{k} a^k b^{p-k} + b^p.$$

For $0 < k < p$, the integer $\binom{p}{k}$ is divisible by $p$; the binomial proof of $a^p \equiv a$ in Fermat's little theorem proves this. Writing $\binom{p}{k} = p\,m$, the term is $m \cdot p\,(a^k b^{p-k}) = 0$. So $(a + b)^p = a^p + b^p$. $\blacksquare$

$\Phi$ is the Frobenius map. On $\mathbb{Z}_p$ it is the identity, which is Fermat's little theorem in the form $a^p = a$. On the field $\mathbb{Z}_3[i]$ with nine elements from Integral Domains and Fields, it is not: since $i^3 = -i$,

$$(a + bi)^3 = a^3 + b^3 i^3 = a - bi,$$

using $a^3 = a$ and $b^3 = b$ in $\mathbb{Z}_3$. So the Frobenius map of $\mathbb{Z}_3[i]$ is conjugation, $a + bi \mapsto a - bi$. It is a bijection and so an automorphism; for example $(1 + i)^3 = 1 + 2i = 1 - i$. The script confirms the formula for all nine elements and both operations for all 81 pairs.

First properties

The examples suggest some rules: every homomorphism so far sent 0 to 0 and negatives to negatives. Do all of them? Here is the idea of the proofs that follow. A homomorphism carries every equation that holds in $R$ to an equation in $S$. Applied to $0 + 0 = 0$ and $a + (-a) = 0$, the cancellation laws of $S$ then force the values of $f(0)$ and $f(-a)$.

Theorem. Let $f: R \to S$ be a homomorphism. Then for all $a, b \in R$:

  1. $f(0_R) = 0_S$;
  2. $f(-a) = -f(a)$ and $f(a - b) = f(a) - f(b)$;
  3. $f(na) = n f(a)$ for every integer $n$, and $f(a^k) = f(a)^k$ for every positive integer $k$;
  4. the image $f(R) = \{f(r) : r \in R\}$ is a subring of $S$, and it is commutative if $R$ is.

Proof. (1) $f(0_R) = f(0_R + 0_R) = f(0_R) + f(0_R)$. Adding $-f(0_R)$ to both sides in $S$ gives $0_S = f(0_R)$.

(2) $f(a) + f(-a) = f(a + (-a)) = f(0_R) = 0_S$ by (1). So $f(-a)$ is an additive inverse of $f(a)$, and additive inverses are unique (Basic Properties of Rings), so $f(-a) = -f(a)$. Then $f(a - b) = f(a + (-b)) = f(a) + f(-b) = f(a) - f(b)$.

(3) For $n = 0$ both sides are $0_S$ by (1). For positive $n$ use induction: $f(1a) = f(a)$, and if $f(na) = nf(a)$, then $f((n + 1)a) = f(na + a) = f(na) + f(a) = nf(a) + f(a) = (n + 1)f(a)$. For negative $n = -m$, $f(na) = f(m(-a)) = mf(-a) = m(-f(a)) = nf(a)$, using the definition $(-m)a = m(-a)$ and (2). Powers are the same induction with products: $f(a^{k+1}) = f(a^k a) = f(a)^k f(a)$.

(4) The image contains $0_S = f(0_R)$. If $f(a)$ and $f(b)$ are in the image, so are $f(a) - f(b) = f(a - b)$ and $f(a) f(b) = f(ab)$. By the subring test, $f(R)$ is a subring. If $R$ is commutative, then $f(a)f(b) = f(ab) = f(ba) = f(b)f(a)$. $\blacksquare$

Parts (1), (2) and (4) are the course text's Theorem 3.10 and Corollary 3.11, and Judson lists (1) and the commutativity statement in Proposition 16.3.3. For the map $g$ above, part (1) is $g(0) = [0]$, and part (2) is $g(-\omega) = [-2] = -g(\omega)$.

The identity

If homomorphisms send 0 to 0, surely they send 1 to 1? That is a natural guess, and it is wrong. The zero map is one counterexample. A less trivial one places a real number in the top left corner of a $2 \times 2$ matrix:

$$f: \mathbb{R} \to M_2(\mathbb{R}), \qquad f(a) = \begin{pmatrix} a & 0 \\ 0 & 0 \end{pmatrix}.$$

  • It respects both operations, because matrices of this shape add and multiply in the corner alone: $f(a)f(b) = \begin{pmatrix} ab & 0 \\ 0 & 0 \end{pmatrix} = f(ab)$, and likewise for sums. It is one-to-one, and the script checks both operations on sample values.
  • It sends 1 to $E = \begin{pmatrix} 1 & 0 \\ 0 & 0 \end{pmatrix}$, not to the identity matrix $I$. And $E$ is not even a unit of $M_2(\mathbb{R})$, since its determinant is 0, although 1 is a unit of $\mathbb{R}$.
  • Its image, the matrices with only a corner entry, is a subring whose own identity is $E$: $E f(a) = f(a) E = f(a)$ for every $a$. Notice that $E^2 = E$.

The Chinese remainder theorem finds elements that behave like $E$ inside $\mathbb{Z}_{45}$.

What can be said is that a homomorphism onto $S$ carries an identity along. The idea is that $1_R$ fixes everything in $R$ under multiplication, so its image fixes everything that is an image, and when $f$ is onto, that is all of $S$.

Proposition. Let $R$ be a ring with identity and $f: R \to S$ a homomorphism.

  1. If $f$ is onto, then $f(1_R)$ is an identity of $S$.
  2. If $S$ has an identity and $f(1_R) = 1_S$, then for every unit $u$ of $R$, $f(u)$ is a unit of $S$ and $f(u^{-1}) = f(u)^{-1}$.

Proof. (1) Let $s \in S$. Since $f$ is onto, $s = f(r)$ for some $r$, and $f(1_R)\,s = f(1_R)f(r) = f(1_R r) = f(r) = s$; in the same way $s\,f(1_R) = s$.

(2) $f(u) f(u^{-1}) = f(uu^{-1}) = f(1_R) = 1_S$, and likewise $f(u^{-1}) f(u) = 1_S$. So $f(u)$ is a unit with inverse $f(u^{-1})$. $\blacksquare$

Part (1) applies to the map $g$ from the Eisenstein integers onto $\mathbb{Z}_7$, which sends 1 to $[1]$, and part (2) is the calculation $g(\omega^2) = g(\omega)^{-1}$ above. The course text states (2) for onto homomorphisms, where (1) supplies $f(1_R) = 1_S$. The corner map above shows that (2) needs its hypothesis: it sends the unit 1 to $E$, which is not a unit.

Homomorphisms out of the integers

Part (3) of the theorem says that a homomorphism out of $\mathbb{Z}$ is decided by where it sends 1. So the integers have a homomorphism into every ring with identity, and only one that sends 1 to 1.

Proposition. Let $R$ be a ring with identity. The map $\chi: \mathbb{Z} \to R$, $\chi(n) = n1_R$, is a homomorphism with $\chi(1) = 1_R$, and it is the only homomorphism from $\mathbb{Z}$ to $R$ with that property.

Proof. Sums: $(m + n)1_R = m1_R + n1_R$ is the first rule for multiples in Basic Properties of Rings. Products: for integers $m$ and $n$ and elements $a$ and $b$, $(ma)b = m(ab)$, by the distributive law and induction on $m$ for $m \ge 0$, and by the rules of signs for negative $m$. Take $a = 1_R$ and $b = n1_R$, and then use the rule $m(na) = (mn)a$ for multiples:

$$(m1_R)(n1_R) = m\big(1_R \cdot n1_R\big) = m(n1_R) = (mn)1_R.$$

For uniqueness, let $h: \mathbb{Z} \to R$ be a homomorphism with $h(1) = 1_R$. By part (3) of the theorem, $h(n) = h(n \cdot 1) = n\,h(1) = n1_R$ for every integer $n$. $\blacksquare$

The image of $\chi$ is the set of multiples of $1_R$, the smallest subring of $R$ that contains $1_R$. Two cases arise.

  • If $n1_R \neq 0$ for every positive $n$, then $\chi$ is one-to-one: if $\chi(a) = \chi(b)$ with $a > b$, then $(a - b)1_R = 0$. So $R$ contains a copy of $\mathbb{Z}$.
  • Otherwise let $c$ be the least positive integer with $c1_R = 0$. Then $a1_R = b1_R$ exactly when $c \mid a - b$. To see this, write $a - b = qc + r$ with $0 \le r < c$ by the division algorithm. Then $(a - b)1_R = q(c1_R) + r1_R = r1_R$, which is 0 exactly when $r = 0$, by the choice of $c$. So $[a]_c \mapsto a1_R$ is a well-defined, one-to-one homomorphism from $\mathbb{Z}_c$ into $R$, and $R$ contains a copy of $\mathbb{Z}_c$.

The number $c$, or 0 in the first case, is the characteristic of $R$. Judson (2026, §16.2) defines it as the least positive $n$ with $na = 0$ for every $a$. He proves that it is the least $n$ with $n1_R = 0$ (Lemma 16.2.7), and that the characteristic of an integral domain is 0 or a prime (Theorem 16.2.8). The script finds that the identity needs 2 additions to reach 0 in the field of four matrices and 3 in the field $\mathbb{Z}_3[i]$, both primes, as Judson's theorem requires. In $\mathbb{Z}[i]$, $n \cdot 1 = n$ is never 0 for positive $n$, so its characteristic is 0. The Frobenius proposition applies to every commutative ring with identity of prime characteristic $p$, since there $pa = (p1_R)a = 0$ for every $a$.

Composites and inverses

Before using isomorphisms to compare rings, we need to know that "is isomorphic to" behaves like a sameness relation. Two facts suffice.

Proposition. Let $f: R \to S$ and $h: S \to T$ be homomorphisms.

  1. The composite $h \circ f: R \to T$ is a homomorphism, and if $f$ and $h$ are isomorphisms, so is $h \circ f$.
  2. If $f$ is an isomorphism, its inverse function $f^{-1}: S \to R$ is an isomorphism.

Proof. (1) $h(f(a + b)) = h(f(a) + f(b)) = h(f(a)) + h(f(b))$, and the same with products. A composite of bijections is a bijection.

(2) $f^{-1}$ is a bijection. Let $s, t \in S$ and put $a = f^{-1}(s)$, $b = f^{-1}(t)$, so $f(a) = s$ and $f(b) = t$. Then $f(a + b) = s + t$, so $f^{-1}(s + t) = a + b = f^{-1}(s) + f^{-1}(t)$. And $f(ab) = st$, so $f^{-1}(st) = ab = f^{-1}(s) f^{-1}(t)$. $\blacksquare$

With the identity map, this shows that "is isomorphic to" is reflexive, symmetric and transitive: an equivalence relation on rings. The classes are the "abstract rings", and a question such as "how many rings with nine elements are there?" means how many classes.

What an isomorphism preserves

An isomorphism matches the elements of two rings one for one and carries every sum and product to the corresponding sum and product. Any property defined only in terms of the elements, the two operations, and counting therefore holds in one ring exactly when it holds in the other. The theorem lists the properties this course uses.

Theorem. Let $f: R \to S$ be an isomorphism. Then:

  1. $R$ and $S$ have the same number of elements;
  2. $R$ is commutative if and only if $S$ is;
  3. $R$ has an identity if and only if $S$ does, and then $f(1_R) = 1_S$;
  4. $u$ is a unit of $R$ if and only if $f(u)$ is a unit of $S$, so $R$ and $S$ have the same number of units;
  5. $a$ is a zero divisor of $R$ if and only if $f(a)$ is a zero divisor of $S$;
  6. $R$ is an integral domain, or a field, if and only if $S$ is;
  7. if $R$ has an identity, $R$ and $S$ have the same characteristic; and for each positive integer $k$, the equation $x^2 = x$, or $x^k = 0$, or (with identities) $x^2 = 1$ has as many solutions in $R$ as in $S$.

Proof. Each statement about $S$ can be pulled back to $R$ with the isomorphism $f^{-1}$, which is an isomorphism by the previous proposition. So it suffices to prove each "only if".

(1) $f$ is a bijection. (2) If $R$ is commutative, so is $f(R) = S$, by part (4) of the first theorem. (3) $f$ is onto, so this is part (1) of the proposition on the identity. (4) By (3), $f(1_R) = 1_S$, and part (2) of the same proposition applies.

(5) If $a \neq 0$ and $ab = 0$ for some $b \neq 0$, then $f(a)f(b) = f(0) = 0$. Both $f(a)$ and $f(b)$ are nonzero, because $f$ is one-to-one and $f(0) = 0$. The case $ba = 0$ is the same.

(6) Being an integral domain or a field is defined by commutativity, an identity different from 0, and the absence of zero divisors or the presence of inverses. These are covered by (2) to (5), and the condition $1 \neq 0$ transfers because $f$ is one-to-one.

(7) $f(n1_R) = nf(1_R) = n1_S$, and $n1_R = 0$ exactly when $f(n1_R) = 0$, since $f$ is one-to-one. And $f$ matches the solutions of each equation in $R$ one for one with those in $S$, because $f(x)^2 = f(x^2)$, $f(x)^k = f(x^k)$ and $f(1_R) = 1_S$. $\blacksquare$

These properties are the tools for proving that two rings are not isomorphic: find one that differs. They never prove the opposite: two rings may agree on size, units, characteristic and every other property one thinks to check, and still not be isomorphic. To show that two rings are isomorphic, an isomorphism has to be produced.

Three rings with nine elements

The rings $\mathbb{Z}_9$, $\mathbb{Z}_3 \times \mathbb{Z}_3$ and $\mathbb{Z}_3[i]$ each have nine elements, and each is commutative with identity. No two are isomorphic:

$\mathbb{Z}_9$ $\mathbb{Z}_3 \times \mathbb{Z}_3$ $\mathbb{Z}_3[i]$
units 6 4 8
field? no no yes
characteristic 9 3 3
nonzero $x$ with $x^2 = 0$ $3, 6$ none none
solutions of $x^2 = 1$ 2 4 2

The counts are the script's. $\mathbb{Z}_3[i]$ is a field (Integral Domains and Fields) and the other two are not, since $3 \cdot 3 = 0$ in $\mathbb{Z}_9$ and $(1, 0)(0, 1) = (0, 0)$ in the product. And $\mathbb{Z}_9 \not\cong \mathbb{Z}_3 \times \mathbb{Z}_3$ for any one of three reasons:

  • they have different numbers of units;
  • they have different characteristics;
  • 3 is a nonzero element of $\mathbb{Z}_9$ with square 0, and the product has none.

Notice that $\mathbb{Z}_3 \times \mathbb{Z}_3$ and $\mathbb{Z}_3[i]$ agree on characteristic and on the absence of nonzero square-zero elements; the units are what separate them.

So there are at least three rings with nine elements up to isomorphism. There are more. For instance, the additive group of $\mathbb{Z}_9$ with every product defined to be 0 satisfies the ring axioms, and it has no identity, so it is isomorphic to none of the three. The comparison of $\mathbb{Z}_9$ with $\mathbb{Z}_3 \times \mathbb{Z}_3$ is a case in which the Chinese remainder theorem does not apply, since 3 and 3 are not relatively prime.

The same tools settle an infinite case. There is no isomorphism from $\mathbb{Z}$ to $\mathbb{Q}$, although both are commutative rings with identity of the same infinite size. An isomorphism would send 1 to 1 and so, by the uniqueness proposition, $n$ to $n$; its image would be the integers inside $\mathbb{Q}$, not all of $\mathbb{Q}$. More briefly, $\mathbb{Q}$ is a field and $\mathbb{Z}$ is not; the course text makes the same comparison by counting units (Hungerford, 2014, §3.3).

Producing an isomorphism

How, then, is an isomorphism found? Write down a candidate map and check it. The properties above constrain the candidate. A map onto $S$ must send $1_R$ to $1_S$, and a homomorphism is determined by its values on elements from which the rest are built by the operations. In $\mathbb{Z}_2[x]/(x^2 + x + 1)$ everything is built from 1 and $\alpha$. So a homomorphism to the matrices with $1 \mapsto I$ has no choice but to send $\alpha$ to a matrix satisfying the same equation $M^2 = M + I$. The search reduces to solving that equation.

Check your understanding. Which of the four matrices $0$, $I$, $A$, $I + A$ satisfy $M^2 = M + I$? How many isomorphisms from $\mathbb{Z}_2[x]/(x^2 + x + 1)$ to the matrices does that allow?

Two: $A$, as in the opening, and $I + A$, since $(I + A)^2 = I + A^2 = I + A + I = A = (I + A) + I$. The matrices $0$ and $I$ fail: $0^2 = 0$ but $0 + I = I$, and $I^2 = I$ but $I + I = 0$. So there are two isomorphisms, $\alpha \mapsto A$ and $\alpha \mapsto I + A$, and the script checks that the second also respects both operations. They differ by the Frobenius map of the field, which sends $\alpha$ to $\alpha^2 = 1 + \alpha$.

Dual numbers. The subrings article found a commutative ring $T$ of real matrices $M(a, b) = \begin{pmatrix} a & b \\ 0 & a \end{pmatrix}$, in which $\varepsilon = M(0, 1)$ satisfies $\varepsilon^2 = 0$ and $M(a, b) = aI + b\varepsilon$. The ring $\mathbb{R}[x]/(x^2)$ consists of the classes $[a + bx]$, one for each pair of reals $a$, $b$ (the remainders of degree less than 2), and its class $[x]$ also squares to 0. Following the recipe, send $[x]$ to $\varepsilon$:

$$\psi\big([a + bx]\big) = M(a, b).$$

Proposition. $\psi$ is an isomorphism from $\mathbb{R}[x]/(x^2)$ onto $T$.

The idea is that both rings multiply pairs $(a, b)$ by the same rule, because in both the second basis element squares to 0.

Proof. Each class has exactly one representative $a + bx$, so $\psi$ is a function. It is a bijection because each $M(a, b)$ comes from exactly one pair $(a, b)$. Sums: $[a + bx] + [c + dx] = [(a + c) + (b + d)x]$ goes to $M(a + c, b + d) = M(a, b) + M(c, d)$. Products: $(a + bx)(c + dx) = ac + (ad + bc)x + bd\,x^2$, and $bd\,x^2$ is a multiple of the modulus, so

$$[a + bx][c + dx] = [ac + (ad + bc)x] \longmapsto M(ac, \, ad + bc) = M(a, b)\,M(c, d),$$

by the product formula for $T$. $\blacksquare$

For instance $(3 + 2x)(5 - x) = 15 + 7x - 2x^2$, whose class is $[15 + 7x]$, and $M(3, 2)M(5, -1) = M(15, 7)$. This ring is the ring of dual numbers. In it, evaluating a real polynomial $f$ at $a + \varepsilon$ gives $f(a) + f'(a)\varepsilon$. By the binomial theorem $(a + \varepsilon)^k = a^k + k a^{k-1} \varepsilon$, since every higher power of $\varepsilon$ is 0, and adding these with the coefficients of $f$ gives $f(a)$ plus the derivative at $a$ times $\varepsilon$. For $f(x) = x^3$ at $a = 2$ this predicts $(2 + \varepsilon)^3 = 8 + 12\varepsilon$, and the script's matrix product $M(2, 1)^3 = M(8, 12)$ agrees. Forward-mode automatic differentiation can be viewed as computing with dual numbers in this way (Baydin et al., 2018, §3.1.1).

The fields $K = \mathbb{Z}_3[x]/(x^2 - 2)$ and $L = \mathbb{Z}_3[x]/(x^2 + x + 2)$ of Building Fields from Irreducible Polynomials were matched the same way, by sending a square root of 2 in $K$ to a square root of 2 in $L$, and the script there checked all 81 pairs. The ring $\mathbb{Z}_3[i]$ is a third copy of that field: $[a + bx] \mapsto a + bi$ is an isomorphism from $\mathbb{Z}_3[x]/(x^2 + 1)$ onto $\mathbb{Z}_3[i]$, by the dual-number proof with $x^2 = -1$ in place of $x^2 = 0$.

Where it leads

What we have learned

A homomorphism respects both operations, and an isomorphism, a bijective homomorphism, shows two rings to be one ring under different names. Every homomorphism sends 0 to 0 and negatives to negatives, and its image is a subring; it need not send 1 to 1. Isomorphisms preserve every property defined by the operations, which told apart three rings with nine elements. Proving that two rings are isomorphic needs an explicit map.

One example points further. The matrix $E$ is its own square, and it acts as an identity on part of the matrix ring but not on all of it. Can a ring as plain as $\mathbb{Z}_{45}$ contain elements like that, and what would they reveal about it? The next article, The Chinese Remainder Theorem, finds two, 36 and 10, and shows that $\mathbb{Z}_{45}$ is $\mathbb{Z}_5 \times \mathbb{Z}_9$ under another name.

References

  • Baydin, A. G., Pearlmutter, B. A., Radul, A. A., and Siskind, J. M. (2018). Automatic differentiation in machine learning: a survey. Journal of Machine Learning Research, 18, 1–43. §3.1.1: forward-mode automatic differentiation viewed as evaluating a function on dual numbers $v + \dot v \varepsilon$ with $\varepsilon^2 = 0$. record
  • Hungerford, T. W. (2014). Abstract Algebra: An Introduction (3rd ed.), §3.3. Brooks/Cole, Cengage Learning. The course text: the definitions of isomorphism and homomorphism; $f(0_R) = 0_S$, $f(-a) = -f(a)$ and $f(a - b) = f(a) - f(b)$, and for onto homomorphisms from a ring with identity, $f(1_R)$ is the identity of $S$ and units go to units (Theorem 3.10); the image of a homomorphism is a subring (Corollary 3.11). record · 44 pages
  • Judson, T. W. (2026). Abstract Algebra: Theory and Applications (annual edition, August 2026). Free online at judsonbooks.org. §16.2: the characteristic of a ring, Lemma 16.2.7 (it is the additive order of 1) and Theorem 16.2.8 (an integral domain has characteristic 0 or a prime); §16.3: ring homomorphisms and isomorphisms, reduction $\mathbb{Z} \to \mathbb{Z}_n$ (Example 16.3.1), evaluation homomorphisms (Example 16.3.2), and first properties (Proposition 16.3.3). record · 44 pages