The Isomorphism Theorems for Rings

How many essentially different quotients can a ring have? Can we list them without doing any computation? This is the concrete question the isomorphism theorems answer completely. Given a ring RR, its ideals correspond exactly to its quotient rings. Every homomorphic image of RR is, up to relabeling, one of these quotients. The isomorphism theorems make this correspondence precise. They turn a vague feeling that "quotienting and mapping are the same operation" into four sharp, provable statements. These theorems matter because they are the load-bearing structure of ring theory. Nearly every later classification result depends on them. This includes the structure of finite fields, the Chinese Remainder Theorem, and the identification of maximal ideals with fields. It also includes localization, and much of algebraic geometry's dictionary between rings and spaces. All of this rests on being able to freely translate between homomorphisms and quotients. Without this dictionary, one would have to re-derive basic facts about images and kernels from scratch every time a new ring appears. The big picture is this. A ring homomorphism φ:R→S\varphi: R \to S carries exactly the same information as a surjection of RR onto a copy of its image, followed by an inclusion. The First Isomorphism Theorem says the image is a quotient of RR by the kernel. The Second and Third refine this. They describe what happens when you build quotients out of sub-objects of sub-objects, or quotients of quotients. The Correspondence Theorem then gives the master picture. The entire ideal lattice of R/IR/I is a faithful, order-preserving snapshot of the ideal lattice of RR lying above II. This article states and proves all four theorems in full, with explicit hypotheses. It includes worked examples with genuinely asymmetric rings, not just Z/nZ\mathbb{Z}/n\mathbb{Z} toy cases. This ensures that every step of the machinery is exercised. It closes by previewing how these results feed directly into the Chinese Remainder Theorem and the classification of Prime Ideals and Maximal Ideals and Fields.

Key takeaways
  • The First Isomorphism Theorem says R/ker⁡φ≅im⁡φR/\ker\varphi \cong \operatorname{im}\varphi for any ring homomorphism φ\varphi.
  • The Second Isomorphism Theorem relates a subring plus an ideal to a quotient built from their sum and intersection.
  • The Third Isomorphism Theorem simplifies "a quotient of a quotient" back to a single quotient: (R/I)/(J/I)≅R/J(R/I)/(J/I) \cong R/J.
  • The Correspondence Theorem shows ideals of R/IR/I biject, order-preservingly, with ideals of RR containing II.
  • All four theorems are proved from the universal property of quotient rings, with no circularity.
  • Together they justify treating "ring homomorphism" and "quotient ring" as two views of one phenomenon.
  • Key idea: every ring homomorphism factors as a quotient map followed by an injection, and this factorization controls all substructure.

Where this lives today

The isomorphism theorems are the machinery behind the Chinese Remainder Theorem, which describes R/(I∩J)R/(I \cap J) in terms of R/IR/I and R/JR/J for comaximal ideals — a direct application of the First Isomorphism Theorem to the product map R→R/I×R/JR \to R/I \times R/J. They are also the tool used to show that a maximal ideal m\mathfrak{m} of a commutative ring RR is equivalent to R/mR/\mathfrak{m} being a field, linking Prime Ideals and Maximal Ideals and Fields to the Correspondence Theorem's picture of the ideal lattice. In algebraic geometry, the Correspondence Theorem underlies the dictionary between subvarieties of an affine variety and ideals of its coordinate ring containing a fixed ideal. In cryptography and coding theory, quotients like Z[x]/(f(x),p)\mathbb{Z}[x]/(f(x), p) (finite fields and their extensions) are analyzed via exactly this factorization of evaluation homomorphisms. In computer algebra systems, Gröbner basis algorithms implicitly use the correspondence between ideals and their quotient rings to compute with polynomial quotient structures efficiently.

:::quiz[In Example 2, why is A∩B={0}A \cap B = \{0\} for A=ZA = \mathbb{Z} and B=(x2−4)B = (x^2-4) in Z[x]\mathbb{Z}[x]?]

  • Because Z\mathbb{Z} and (x2−4)(x^2-4) are comaximal
  • Because any nonzero multiple of x2−4x^2-4 has degree at least 2, so cannot equal a nonzero constant ✓
  • Because x2−4x^2 - 4 has integer roots ±2\pm 2
  • Because AA is not actually a subring of Z[x]\mathbb{Z}[x]

Ask the tutor: How do the isomorphism theorems specialize to prove the Chinese Remainder Theorem for Z/mnZ\mathbb{Z}/mn\mathbb{Z} when gcd⁡(m,n)=1\gcd(m,n)=1?

DefinitionRing homomorphisms, kernels, and quotient rings (recalled)

Let RR and SS be rings (not necessarily commutative, not assumed to have identity unless stated; we use "ring" to mean a set with +,⋅+,\cdot satisfying the usual axioms). A ring homomorphism φ:R→S\varphi: R \to S is a map satisfying

φ(a+b)=φ(a)+φ(b),φ(ab)=φ(a)φ(b)\varphi(a+b) = \varphi(a)+\varphi(b), \qquad \varphi(ab) = \varphi(a)\varphi(b)

for all a,b∈Ra,b \in R (and φ(1R)=1S\varphi(1_R) = 1_S if both rings have identity and the homomorphism is required to be unital). The kernel is

ker⁡φ={a∈R:φ(a)=0S},\ker\varphi = \{a \in R : \varphi(a) = 0_S\},

which is always a (two-sided) ideal of RR: an additive subgroup I≤RI \le R satisfying rI⊆IrI \subseteq I and Ir⊆IIr \subseteq I for all r∈Rr \in R. Given an ideal I⊴RI \trianglelefteq R, the quotient ring R/IR/I is the set of cosets {a+I:a∈R}\{a + I : a \in R\} with operations

(a+I)+(b+I)=(a+b)+I,(a+I)(b+I)=ab+I,(a+I)+(b+I) = (a+b)+I, \qquad (a+I)(b+I) = ab+I,

which are well-defined precisely because II is an ideal. The canonical projection π:R→R/I\pi: R \to R/I, π(a)=a+I\pi(a) = a+I, is a surjective ring homomorphism with ker⁡π=I\ker\pi = I.

Intuition

Think of a ring homomorphism φ:R→S\varphi: R \to S as a lossy compression of RR. Elements that get sent to the same value in SS are, from the point of view of SS, indistinguishable. The kernel ker⁡φ\ker\varphi collects exactly the elements that get squashed to 00, and it turns out that two elements a,ba, b map to the same place in SS if and only if a−b∈ker⁡φa - b \in \ker\varphi. So the partition of RR into "fibers of φ\varphi" is exactly the partition of RR into cosets of ker⁡φ\ker\varphi. The First Isomorphism Theorem is the statement that this coset partition, equipped with its natural ring structure, is not just in bijection with the image but is a ring isomorphic copy of it. The Second and Third theorems are bookkeeping devices for combining and simplifying quotients. If you have already quotiented by II, and you want to quotient further by an ideal that lives "above" II, the Third Isomorphism Theorem tells you this two-step quotient collapses to a single quotient by a bigger ideal — nothing is lost or gained by doing it in stages. The Correspondence Theorem is the payoff: it says the entire "ideal skeleton" of R/IR/I is just a relabeled copy of the part of the ideal skeleton of RR sitting above II, with nothing added or removed.

Figure
Lattice of ideals of Z\mathbb{Z} containing 12Z12\mathbb{Z}, corresponding to ideals of Z/12Z\mathbb{Z}/12\mathbb{Z}
ℤ 2ℤ 3ℤ ?ℤ 4ℤ 6ℤ 3ℤ' 12ℤ Ideals of ℤ above 12ℤ ⟷ ideals of ℤ/12ℤ = {0̄,2̄,3̄,4̄,6̄,1̄}
TheoremFirst Isomorphism Theorem for Rings

Let RR and SS be rings and let φ:R→S\varphi: R \to S be a ring homomorphism. Then: 1. ker⁡φ\ker\varphi is an ideal of RR. 2. im⁡φ=φ(R)\operatorname{im}\varphi = \varphi(R) is a subring of SS. 3. The map

φˉ:R/ker⁡φ→im⁡φ,φˉ(a+ker⁡φ)=φ(a)\bar\varphi: R/\ker\varphi \to \operatorname{im}\varphi, \qquad \bar\varphi(a+\ker\varphi) = \varphi(a)

is a well-defined ring isomorphism. Consequently,

R/ker⁡φ  ≅  im⁡φ.R/\ker\varphi \;\cong\; \operatorname{im}\varphi.
Proof

Write K=ker⁡φK = \ker\varphi.

(1) KK is an ideal. Since φ(0R)=φ(0R+0R)=φ(0R)+φ(0R)\varphi(0_R) = \varphi(0_R + 0_R) = \varphi(0_R)+\varphi(0_R), subtracting φ(0R)\varphi(0_R) from both sides in the abelian group (S,+)(S,+) gives φ(0R)=0S\varphi(0_R) = 0_S, so 0R∈K0_R \in K. If a,b∈Ka,b \in K then φ(a−b)=φ(a)−φ(b)=0S−0S=0S\varphi(a-b) = \varphi(a) - \varphi(b) = 0_S - 0_S = 0_S (using that φ\varphi is additive, hence a group homomorphism on (R,+)(R,+), so φ(−b)=−φ(b)\varphi(-b) = -\varphi(b)); thus a−b∈Ka - b \in K, so KK is an additive subgroup. For r∈Rr \in R and a∈Ka \in K, φ(ra)=φ(r)φ(a)=φ(r)⋅0S=0S\varphi(ra) = \varphi(r)\varphi(a) = \varphi(r)\cdot 0_S = 0_S, so ra∈Kra \in K; similarly φ(ar)=φ(a)φ(r)=0S\varphi(ar) = \varphi(a)\varphi(r) = 0_S, so ar∈Kar \in K. Hence KK absorbs multiplication by RR on both sides, so KK is a two-sided ideal.

(2) im⁡φ\operatorname{im}\varphi is a subring. It is nonempty (0S∈im⁡φ0_S \in \operatorname{im}\varphi). If φ(a),φ(b)∈im⁡φ\varphi(a), \varphi(b) \in \operatorname{im}\varphi then φ(a)−φ(b)=φ(a−b)∈im⁡φ\varphi(a) - \varphi(b) = \varphi(a-b) \in \operatorname{im}\varphi and φ(a)φ(b)=φ(ab)∈im⁡φ\varphi(a)\varphi(b) = \varphi(ab) \in \operatorname{im}\varphi. So im⁡φ\operatorname{im}\varphi is closed under subtraction and multiplication, hence a subring of SS.

(3) Well-definedness of φˉ\bar\varphi. Suppose a+K=a′+Ka + K = a' + K in R/KR/K, i.e. a−a′∈Ka - a' \in K. Then φ(a)−φ(a′)=φ(a−a′)=0S\varphi(a) - \varphi(a') = \varphi(a-a') = 0_S, so φ(a)=φ(a′)\varphi(a) = \varphi(a'). Hence φˉ(a+K):=φ(a)\bar\varphi(a+K) := \varphi(a) does not depend on the choice of coset representative, and φˉ\bar\varphi is a well-defined function R/K→im⁡φR/K \to \operatorname{im}\varphi.

φˉ\bar\varphi is a ring homomorphism. For a,b∈Ra,b \in R:

φˉ((a+K)+(b+K))=φˉ((a+b)+K)=φ(a+b)=φ(a)+φ(b)=φˉ(a+K)+φˉ(b+K),\bar\varphi\big((a+K)+(b+K)\big) = \bar\varphi\big((a+b)+K\big) = \varphi(a+b) = \varphi(a)+\varphi(b) = \bar\varphi(a+K)+\bar\varphi(b+K),
φˉ((a+K)(b+K))=φˉ(ab+K)=φ(ab)=φ(a)φ(b)=φˉ(a+K) φˉ(b+K).\bar\varphi\big((a+K)(b+K)\big) = \bar\varphi\big(ab+K\big) = \varphi(ab) = \varphi(a)\varphi(b) = \bar\varphi(a+K)\,\bar\varphi(b+K).

These use exactly the definitions of addition and multiplication in R/KR/K from Quotient Rings, which are well-defined because KK is an ideal by part (1).

φˉ\bar\varphi is surjective. Any element of im⁡φ\operatorname{im}\varphi has the form φ(a)\varphi(a) for some a∈Ra \in R, and φˉ(a+K)=φ(a)\bar\varphi(a+K) = \varphi(a), so φˉ\bar\varphi hits every element of im⁡φ\operatorname{im}\varphi.

φˉ\bar\varphi is injective. Suppose φˉ(a+K)=φˉ(b+K)\bar\varphi(a+K) = \bar\varphi(b+K), i.e. φ(a)=φ(b)\varphi(a) = \varphi(b). Then φ(a−b)=φ(a)−φ(b)=0S\varphi(a-b) = \varphi(a)-\varphi(b) = 0_S, so a−b∈Ka - b \in K, which means a+K=b+Ka+K = b+K. So φˉ\bar\varphi is injective. Since φˉ\bar\varphi is a bijective ring homomorphism, it is a ring isomorphism, and R/K≅im⁡φR/K \cong \operatorname{im}\varphi. ■\blacksquare

∎
LemmaSum and intersection of a subring and an ideal

Let RR be a ring, let A≤RA \le R be a subring, and let B⊴RB \trianglelefteq R be an ideal. Then: 1. A+B={a+b:a∈A,b∈B}A + B = \{a+b : a\in A, b\in B\} is a subring of RR. 2. BB is an ideal of A+BA+B. 3. A∩BA \cap B is an ideal of AA.

Proof

(1) A+BA+B contains 0=0+00 = 0+0. If a1+b1,a2+b2∈A+Ba_1+b_1, a_2+b_2 \in A+B (with ai∈A,bi∈Ba_i \in A, b_i \in B), then

(a1+b1)−(a2+b2)=(a1−a2)+(b1−b2)∈A+B(a_1+b_1)-(a_2+b_2) = (a_1-a_2)+(b_1-b_2) \in A+B

since AA is closed under subtraction and BB is closed under subtraction. For products,

(a1+b1)(a2+b2)=a1a2+a1b2+b1a2+b1b2.(a_1+b_1)(a_2+b_2) = a_1a_2 + a_1b_2 + b_1a_2 + b_1b_2.

Here a1a2∈Aa_1a_2 \in A since AA is a subring; and a1b2,b1a2,b1b2∈Ba_1 b_2, b_1 a_2, b_1 b_2 \in B since BB is an ideal (absorbing multiplication by any element of RR, in particular by elements of A⊆RA \subseteq R), and BB is closed under addition. So the whole sum lies in A+BA + B (write it as the sum a1a2∈Aa_1a_2 \in A plus the ideal element a1b2+b1a2+b1b2∈Ba_1b_2+b_1a_2+b_1b_2 \in B). Hence A+BA+B is closed under subtraction and multiplication, so it is a subring of RR.

(2) B⊆A+BB \subseteq A+B and BB is an additive subgroup, hence an additive subgroup of A+BA+B. Since B⊴RB \trianglelefteq R and A+B⊆RA + B \subseteq R, for any x∈A+B⊆Rx \in A+B \subseteq R we have xB⊆BxB \subseteq B and Bx⊆BBx \subseteq B. So BB is an ideal of A+BA+B.

(3) A∩BA \cap B is an intersection of two additive subgroups of RR, hence an additive subgroup of AA (it is a subset of AA closed under the group operations). For a∈Aa \in A and x∈A∩Bx \in A \cap B: ax∈Aax \in A (since AA is a subring and a,x∈Aa, x \in A) and ax∈Bax \in B (since x∈Bx \in B and B⊴RB \trianglelefteq R absorbs multiplication by a∈Ra \in R); so ax∈A∩Bax \in A \cap B. Similarly xa∈A∩Bxa \in A \cap B. Hence A∩BA \cap B is an ideal of AA. ■\blacksquare

∎
TheoremSecond Isomorphism Theorem for Rings (Diamond Isomorphism Theorem)

Let RR be a ring, let A≤RA \le R be a subring, and let B⊴RB \trianglelefteq R be an ideal. Then A+BA+B is a subring of RR, BB is an ideal of A+BA+B, A∩BA \cap B is an ideal of AA, and there is a ring isomorphism

(A+B)/B  ≅  A/(A∩B).(A+B)/B \;\cong\; A/(A\cap B).
Proof

The subring/ideal claims are exactly the preceding Sum and intersection of a subring and an ideal lemma, so it remains to construct the isomorphism. Define φ:A→(A+B)/B\varphi: A \to (A+B)/B by

φ(a)=a+B.\varphi(a) = a + B.

This makes sense because a∈A⊆A+Ba \in A \subseteq A+B, so a+Ba+B is a genuine coset in (A+B)/B(A+B)/B.

φ\varphi is a ring homomorphism. For a1,a2∈Aa_1, a_2 \in A,

φ(a1+a2)=(a1+a2)+B=(a1+B)+(a2+B)=φ(a1)+φ(a2),\varphi(a_1+a_2) = (a_1+a_2)+B = (a_1+B)+(a_2+B) = \varphi(a_1)+\varphi(a_2),
φ(a1a2)=a1a2+B=(a1+B)(a2+B)=φ(a1)φ(a2),\varphi(a_1a_2) = a_1a_2+B = (a_1+B)(a_2+B) = \varphi(a_1)\varphi(a_2),

using the definitions of addition and multiplication on (A+B)/B(A+B)/B (valid since B⊴A+BB \trianglelefteq A+B by the lemma).

φ\varphi is surjective. An arbitrary element of (A+B)/B(A+B)/B has the form (a+b)+B(a+b)+B for some a∈A,b∈Ba \in A, b \in B. Since b∈Bb \in B, (a+b)−a=b∈B(a+b) - a = b \in B, so (a+b)+B=a+B=φ(a)(a+b)+B = a+B = \varphi(a). Hence every coset in (A+B)/B(A+B)/B is hit.

Computing the kernel. We claim ker⁡φ=A∩B\ker\varphi = A \cap B. If a∈Aa \in A and φ(a)=0+B\varphi(a) = 0 + B (the zero coset), then a∈Ba \in B by definition of the coset equivalence (a−0=a∈Ba - 0 = a \in B); combined with a∈Aa \in A this gives a∈A∩Ba \in A\cap B. Conversely if a∈A∩Ba \in A \cap B then a∈Ba \in B, so a+B=0+Ba + B = 0 + B, i.e. φ(a)=0\varphi(a) = 0 in (A+B)/B(A+B)/B. So ker⁡φ=A∩B\ker\varphi = A \cap B.

Conclusion. By the First Isomorphism Theorem applied to φ:A→(A+B)/B\varphi: A \to (A+B)/B, which is surjective with kernel A∩BA \cap B,

A/(A∩B)  ≅  im⁡φ=(A+B)/B.A/(A\cap B) \;\cong\; \operatorname{im}\varphi = (A+B)/B.

This is the desired isomorphism (stated in the other order by convention). ■\blacksquare

∎
TheoremThird Isomorphism Theorem for Rings

Let RR be a ring and let I,JI, J be ideals of RR with I⊆JI \subseteq J. Then J/IJ/I is an ideal of R/IR/I, and there is a ring isomorphism

(R/I)/(J/I)  ≅  R/J.(R/I)\big/(J/I) \;\cong\; R/J.
Proof

Step 1: J/IJ/I makes sense and is an ideal of R/IR/I. Since I⊆JI \subseteq J, every coset j+Ij + I with j∈Jj \in J is a well-defined element of R/IR/I; set J/I:={j+I:j∈J}⊆R/IJ/I := \{j+I : j \in J\} \subseteq R/I. This is an additive subgroup of R/IR/I (image of the additive subgroup JJ under the additive group homomorphism π:R→R/I\pi: R \to R/I, π(r)=r+I\pi(r)=r+I, restricted to JJ). For r+I∈R/Ir+I \in R/I and j+I∈J/Ij+I \in J/I: (r+I)(j+I)=rj+I(r+I)(j+I) = rj + I, and rj∈Jrj \in J because J⊴RJ \trianglelefteq R; so rj+I∈J/Irj + I \in J/I. Similarly jr+I∈J/Ijr+I \in J/I. Hence J/I⊴R/IJ/I \trianglelefteq R/I, and the quotient ring (R/I)/(J/I)(R/I)/(J/I) is defined.

Step 2: Construct a surjective homomorphism R→(R/I)/(J/I)R \to (R/I)/(J/I) with kernel JJ. Let π1:R→R/I\pi_1: R \to R/I be the canonical projection π1(r)=r+I\pi_1(r) = r+I, and let π2:R/I→(R/I)/(J/I)\pi_2: R/I \to (R/I)/(J/I) be the canonical projection π2(x)=x+(J/I)\pi_2(x) = x + (J/I). Both are surjective ring homomorphisms (canonical projections onto quotient rings are always surjective ring homomorphisms, as recalled in the definition block). Let

ψ=π2∘π1:R→(R/I)/(J/I),ψ(r)=(r+I)+(J/I).\psi = \pi_2 \circ \pi_1 : R \to (R/I)/(J/I), \qquad \psi(r) = (r+I) + (J/I).

As a composite of ring homomorphisms, ψ\psi is a ring homomorphism. As a composite of surjections, ψ\psi is surjective.

Step 3: Compute ker⁡ψ=J\ker\psi = J. We have ψ(r)=0\psi(r) = 0 (the zero coset of (R/I)/(J/I)(R/I)/(J/I), i.e. the coset J/IJ/I itself) if and only if π1(r)=r+I∈J/I\pi_1(r) = r+I \in J/I, i.e. if and only if r+I=j+Ir + I = j + I for some j∈Jj \in J, i.e. if and only if r−j∈I⊆Jr - j \in I \subseteq J for some j∈Jj \in J. Since j∈Jj \in J and r−j∈Jr - j \in J (as I⊆JI \subseteq J), adding gives r=(r−j)+j∈Jr = (r-j) + j \in J. Conversely if r∈Jr \in J then r+I∈J/Ir + I \in J/I trivially (take j=rj = r), so ψ(r)=0\psi(r) = 0. Hence

ker⁡ψ=J.\ker\psi = J.

Step 4: Apply the First Isomorphism Theorem. Since ψ:R→(R/I)/(J/I)\psi: R \to (R/I)/(J/I) is a surjective ring homomorphism with ker⁡ψ=J\ker\psi = J, the First Isomorphism Theorem gives

R/J  ≅  (R/I)/(J/I).R/J \;\cong\; (R/I)/(J/I).

■\blacksquare

∎
TheoremCorrespondence Theorem for Rings (Lattice Isomorphism Theorem)

Let RR be a ring and let I⊴RI \trianglelefteq R be an ideal. Let π:R→R/I\pi : R \to R/I be the canonical projection. Then π\pi induces an inclusion-preserving bijection

{ ideals J of R with I⊆J⊆R }  ⟷  { ideals of R/I }\{\, \text{ideals } J \text{ of } R \text{ with } I \subseteq J \subseteq R \,\} \;\longleftrightarrow\; \{\, \text{ideals of } R/I \,\}

given by J↦J/I=π(J)J \mapsto J/I = \pi(J), with inverse given by Jˉ↦π−1(Jˉ)\bar J \mapsto \pi^{-1}(\bar J). Moreover this bijection preserves inclusions in both directions (J1⊆J2  ⟺  J1/I⊆J2/IJ_1 \subseteq J_2 \iff J_1/I \subseteq J_2/I), and for such JJ,

(R/I)/(J/I)≅R/J.(R/I)/(J/I) \cong R/J.

The same statement holds with "ideal" replaced by "subring."

Proof

Let JI\mathcal{J}_I denote the set of ideals of RR containing II, and let Q\mathcal{Q} denote the set of ideals of R/IR/I. Define

Φ:JI→Q,Φ(J)=π(J)=J/I,Ψ:Q→JI,Ψ(Jˉ)=π−1(Jˉ).\Phi: \mathcal{J}_I \to \mathcal{Q}, \quad \Phi(J) = \pi(J) = J/I, \qquad \Psi: \mathcal{Q} \to \mathcal{J}_I, \quad \Psi(\bar J) = \pi^{-1}(\bar J).

Φ\Phi is well-defined, i.e. π(J)\pi(J) is an ideal of R/IR/I for J∈JIJ \in \mathcal{J}_I. Since I⊆JI \subseteq J, this was verified in Step 1 of the Third Isomorphism Theorem proof.

Ψ\Psi is well-defined, i.e. π−1(Jˉ)\pi^{-1}(\bar J) is an ideal of RR containing II, for any ideal Jˉ⊴R/I\bar J \trianglelefteq R/I. Preimages of ideals under ring homomorphisms are always ideals: if x,y∈π−1(Jˉ)x, y \in \pi^{-1}(\bar J) then π(x−y)=π(x)−π(y)∈Jˉ\pi(x-y) = \pi(x)-\pi(y) \in \bar J (as Jˉ\bar J is an additive subgroup), so x−y∈π−1(Jˉ)x - y \in \pi^{-1}(\bar J); and for r∈Rr \in R, π(rx)=π(r)π(x)∈Jˉ\pi(rx) = \pi(r)\pi(x) \in \bar J (as Jˉ\bar J absorbs multiplication by the arbitrary element π(r)∈R/I\pi(r) \in R/I), so rx∈π−1(Jˉ)rx \in \pi^{-1}(\bar J), and similarly xr∈π−1(Jˉ)xr \in \pi^{-1}(\bar J). Also π−1(Jˉ)⊇π−1({0})=ker⁡π=I\pi^{-1}(\bar J) \supseteq \pi^{-1}(\{0\}) = \ker\pi = I because 0∈Jˉ0 \in \bar J. So Ψ(Jˉ)∈JI\Psi(\bar J) \in \mathcal{J}_I.

Ψ∘Φ=id\Psi \circ \Phi = \mathrm{id}. Let J∈JIJ \in \mathcal{J}_I. We must show π−1(π(J))=J\pi^{-1}(\pi(J)) = J. The inclusion J⊆π−1(π(J))J \subseteq \pi^{-1}(\pi(J)) is immediate (if x∈Jx \in J then π(x)∈π(J)\pi(x) \in \pi(J), so x∈π−1(π(J))x \in \pi^{-1}(\pi(J))). For the reverse inclusion, suppose x∈π−1(π(J))x \in \pi^{-1}(\pi(J)), i.e. π(x)=π(j)\pi(x) = \pi(j) for some j∈Jj \in J. Then x−j∈ker⁡π=I⊆Jx - j \in \ker\pi = I \subseteq J (using the hypothesis I⊆JI \subseteq J). Since j∈Jj \in J and x−j∈Jx - j \in J, we get x=(x−j)+j∈Jx = (x-j)+j \in J. Hence π−1(π(J))⊆J\pi^{-1}(\pi(J)) \subseteq J, giving equality. This step crucially uses I⊆JI \subseteq J; without it, the argument fails (see the counterexample below).

Φ∘Ψ=id\Phi \circ \Psi = \mathrm{id}. Let Jˉ∈Q\bar J \in \mathcal{Q}. We must show π(π−1(Jˉ))=Jˉ\pi(\pi^{-1}(\bar J)) = \bar J. The inclusion π(π−1(Jˉ))⊆Jˉ\pi(\pi^{-1}(\bar J)) \subseteq \bar J holds for any function's preimage. For the reverse inclusion, since π\pi is surjective, every element of Jˉ⊆R/I\bar J \subseteq R/I has the form π(x)\pi(x) for some x∈Rx \in R; then x∈π−1(Jˉ)x \in \pi^{-1}(\bar J) by definition, so π(x)∈π(π−1(Jˉ))\pi(x) \in \pi(\pi^{-1}(\bar J)). This uses surjectivity of π\pi, which holds since π\pi is a canonical projection. Thus Φ\Phi and Ψ\Psi are mutually inverse bijections between JI\mathcal{J}_I and Q\mathcal{Q}.

Order preservation. If J1⊆J2J_1 \subseteq J_2 in JI\mathcal{J}_I, then clearly π(J1)⊆π(J2)\pi(J_1) \subseteq \pi(J_2), i.e. Φ(J1)⊆Φ(J2)\Phi(J_1) \subseteq \Phi(J_2), since π\pi is a function. Conversely, if Φ(J1)⊆Φ(J2)\Phi(J_1) \subseteq \Phi(J_2), apply Ψ\Psi (order-preserving by the same reasoning, since preimages of nested sets are nested) to get J1=ΨΦ(J1)⊆ΨΦ(J2)=J2J_1 = \Psi\Phi(J_1) \subseteq \Psi\Phi(J_2) = J_2, using the bijectivity just proved. So J1⊆J2  ⟺  Φ(J1)⊆Φ(J2)J_1 \subseteq J_2 \iff \Phi(J_1) \subseteq \Phi(J_2).

The quotient formula. For J∈JIJ \in \mathcal{J}_I, i.e. I⊆JI \subseteq J, the isomorphism (R/I)/(J/I)≅R/J(R/I)/(J/I) \cong R/J is exactly the Third Isomorphism Theorem, already proved above, and J/I=Φ(J)J/I = \Phi(J) by definition.

The subring version. The identical argument works verbatim replacing "ideal" by "subring" throughout: preimages and images of subrings under a ring homomorphism are subrings (the same closure-under-subtraction-and-multiplication computations apply, using only that π(x−y)=π(x)−π(y)\pi(x-y)=\pi(x)-\pi(y) and π(xy)=π(x)π(y)\pi(xy)=\pi(x)\pi(y), without ever using the ideal-absorption property), and the same two inclusion arguments (J⊆π−1π(J)J \subseteq \pi^{-1}\pi(J) using I⊆JI \subseteq J, and ππ−1(Jˉ)=Jˉ\pi\pi^{-1}(\bar J) = \bar J using surjectivity) go through unchanged. ■\blacksquare

∎

Worked examples

Example 1: First and Third Isomorphism Theorems in Z[x]\mathbb{Z}[x]. Let R=Z[x]R = \mathbb{Z}[x] and let φ:Z[x]→Z/6Z\varphi: \mathbb{Z}[x] \to \mathbb{Z}/6\mathbb{Z} be evaluation at 22 followed by reduction mod 66, i.e.

φ(f(x))=f(2)  6.\varphi(f(x)) = f(2) \bmod 6.

This is a ring homomorphism because evaluation at a fixed element followed by a ring homomorphism (Z→Z/6Z\mathbb{Z} \to \mathbb{Z}/6\mathbb{Z}) is a composite of ring homomorphisms. Kernel. We claim ker⁡φ=(x−2,6)\ker\varphi = (x-2, 6), the ideal generated by x−2x-2 and 66. Indeed, f(x)∈ker⁡φf(x) \in \ker\varphi means f(2)≡0(mod6)f(2) \equiv 0 \pmod 6. Writing f(x)=(x−2)q(x)+f(2)f(x) = (x-2)q(x) + f(2) (polynomial division by the monic linear polynomial x−2x-2), we see f(x)∈(x−2,6)f(x) \in (x-2,6) if and only if f(2)∈6Zf(2) \in 6\mathbb{Z}, i.e. iff φ(f)=0\varphi(f)=0. So indeed ker⁡φ=(x−2,6)\ker\varphi = (x-2,6). Image. φ\varphi is surjective onto Z/6Z\mathbb{Z}/6\mathbb{Z}: for any nˉ∈Z/6Z\bar{n} \in \mathbb{Z}/6\mathbb{Z}, the constant polynomial nn maps to nˉ\bar n. By the First Isomorphism Theorem,

Z[x]/(x−2,6)  ≅  Z/6Z.\mathbb{Z}[x]/(x-2,6) \;\cong\; \mathbb{Z}/6\mathbb{Z}.

Now use the Third Isomorphism Theorem to see this as a two-step quotient. Let I=(x−2)⊴Z[x]I = (x-2) \trianglelefteq \mathbb{Z}[x]. There is an isomorphism Z[x]/(x−2)≅Z\mathbb{Z}[x]/(x-2) \cong \mathbb{Z} (evaluation at 22, an entirely analogous First Isomorphism Theorem argument with kernel exactly (x−2)(x-2) since f(2)=0  ⟺  (x−2)∣f(x)f(2)=0 \iff (x-2)\mid f(x) in Z[x]\mathbb{Z}[x], valid because x−2x-2 is monic). Under this isomorphism, the ideal (x−2,6)/(x−2)(x-2,6)/(x-2) of Z[x]/(x−2)\mathbb{Z}[x]/(x-2) corresponds to the ideal 6Z6\mathbb{Z} of Z\mathbb{Z} (both generated by the image of 66). Taking I=(x−2)⊆J=(x−2,6)I = (x-2) \subseteq J = (x-2,6) in the Third Isomorphism Theorem:

(Z[x]/(x−2))/((x−2,6)/(x−2))  ≅  Z[x]/(x−2,6),\big(\mathbb{Z}[x]/(x-2)\big)\big/\big((x-2,6)/(x-2)\big) \;\cong\; \mathbb{Z}[x]/(x-2,6),

and translating the left side via Z[x]/(x−2)≅Z\mathbb{Z}[x]/(x-2)\cong \mathbb{Z} gives Z/6Z≅Z[x]/(x−2,6)\mathbb{Z}/6\mathbb{Z} \cong \mathbb{Z}[x]/(x-2,6) — matching the direct First Isomorphism Theorem computation above, exactly as the theorem predicts.

Example 2: Second Isomorphism Theorem with genuinely asymmetric AA and BB. Let R=Z[x]R = \mathbb{Z}[x]. Let A=ZA = \mathbb{Z} (constant polynomials), a subring but not an ideal of Z[x]\mathbb{Z}[x] (e.g. x⋅1=x∉Zx \cdot 1 = x \notin \mathbb{Z}). Let B=(x2−4)=(x−2)(x+2)B = (x^2 - 4) = (x-2)(x+2), the principal ideal generated by x2−4x^2-4. Computing A∩BA \cap B. An element of A∩BA \cap B is an integer nn that is also a multiple of x2−4x^2-4 in Z[x]\mathbb{Z}[x], say n=(x2−4)g(x)n = (x^2-4)g(x). If g≠0g \neq 0 then deg⁡((x2−4)g(x))≥2>0=deg⁡(n)\deg\big((x^2-4)g(x)\big) \ge 2 > 0 = \deg(n) (for n≠0n \ne 0), a contradiction unless g=0g = 0, forcing n=0n = 0. So

A∩B={0}.A \cap B = \{0\}.

Computing A+BA + B. By definition A+B={n+(x2−4)g(x):n∈Z, g∈Z[x]}A+B = \{n + (x^2-4)g(x) : n \in \mathbb{Z},\, g \in \mathbb{Z}[x]\}, the set of all integer translates of multiples of x2−4x^2-4. By the Second Isomorphism Theorem,

(A+B)/B  ≅  A/(A∩B)=Z/{0}≅Z.(A+B)/B \;\cong\; A/(A\cap B) = \mathbb{Z}/\{0\} \cong \mathbb{Z}.

This says: inside Z[x]/(x2−4)\mathbb{Z}[x]/(x^2-4), the image of the constant subring Z\mathbb{Z} (namely (A+B)/B(A+B)/B) is isomorphic to Z\mathbb{Z} itself — the map Z→Z[x]/(x2−4)\mathbb{Z} \to \mathbb{Z}[x]/(x^2-4) sending n↦n+(x2−4)n \mapsto n + (x^2-4) is injective. This is a nontrivial and useful fact: it confirms that no nonzero integer becomes "identified with 00" after quotienting by (x2−4)(x^2-4), something not obvious without the theorem, since a priori a large enough combination might have collapsed. We can double check directly: if n+(x2−4)=0+(x2−4)n + (x^2-4) = 0 + (x^2-4) for an integer nn, then n∈(x2−4)n \in (x^2-4), forcing n=0n = 0 by the degree argument above — confirming injectivity and matching the isomorphism theorem's conclusion.

CounterexampleDropping the hypothesis I⊆JI \subseteq J in the Correspondence Theorem

The Correspondence Theorem crucially requires I⊆JI \subseteq J before matching ideals of RR containing II with ideals of R/IR/I. Drop this and the map J↦π(J)J \mapsto \pi(J) can fail to be injective, wrecking the bijection. Take R=ZR = \mathbb{Z}, I=6ZI = 6\mathbb{Z}, and let π:Z→Z/6Z\pi: \mathbb{Z} \to \mathbb{Z}/6\mathbb{Z} be the canonical projection. Consider two ideals of RR that do not contain II: J1=2ZJ_1 = 2\mathbb{Z} actually does contain II (since 6Z⊆2Z6\mathbb{Z}\subseteq 2\mathbb{Z}), so instead pick ideals not containing II, e.g. J1=4ZJ_1 = 4\mathbb{Z} and J2=10ZJ_2 = 10\mathbb{Z}; neither contains 6Z6\mathbb{Z} (e.g. 6∉4Z6 \notin 4\mathbb{Z} and 6∉10Z6 \notin 10\mathbb{Z}). Now compute images under π\pi. We have J1=4Z={…,−4,0,4,8,… }J_1 = 4\mathbb{Z} = \{\dots,-4,0,4,8,\dots\}, and reducing mod 66:

π(J1)={0‾,4‾,2‾}={0,2,4}⊆Z/6Z,\pi(J_1) = \{\overline{0}, \overline{4}, \overline{2}\} = \{0,2,4\} \subseteq \mathbb{Z}/6\mathbb{Z},

since 4⋅0=0,4⋅1=4,4⋅2=8≡2,4⋅3=12≡0,…4\cdot 0=0, 4\cdot1=4, 4\cdot2=8\equiv2, 4\cdot3=12\equiv0,\dots cycling through {0,4,2}\{0,4,2\}. Similarly J2=10ZJ_2 = 10\mathbb{Z} reduces mod 6: 10≡410 \equiv 4, so π(J2)\pi(J_2) is generated by 4‾\overline{4} in Z/6Z\mathbb{Z}/6\mathbb{Z}, giving the same set {0,2,4}\{0,2,4\}. So π(J1)=π(J2)={0,2,4}\pi(J_1) = \pi(J_2) = \{0,2,4\} even though J1=4Z≠10Z=J2J_1 = 4\mathbb{Z} \ne 10\mathbb{Z} = J_2 as ideals of Z\mathbb{Z}. The map J↦π(J)J \mapsto \pi(J) is not injective once we allow ideals not containing ker⁡π=I\ker\pi = I, so the bijective correspondence fails without the hypothesis I⊆JI \subseteq J. The proof breaks exactly at the step showing π−1(π(J))=J\pi^{-1}(\pi(J)) = J: for J=4ZJ = 4\mathbb{Z}, in fact π−1(π(J))=π−1({0,2,4})=2Z⊋4Z\pi^{-1}(\pi(J)) = \pi^{-1}(\{0,2,4\}) = 2\mathbb{Z} \supsetneq 4\mathbb{Z}, so ΨΦ(J)≠J\Psi\Phi(J) \ne J, precisely because I=6Z⊈4Z=JI = 6\mathbb{Z} \not\subseteq 4\mathbb{Z} = J.

Scope, edge cases & common traps

SituationWhat holdsWhat can fail
φ\varphi not surjectiveFirst Isom. Thm. still gives R/ker⁡φ≅im⁡φR/\ker\varphi \cong \operatorname{im}\varphi (image is a proper subring of codomain)R/ker⁡φ≅SR/\ker\varphi \cong S fails if φ\varphi isn't onto SS
Rings without unity, or non-unital homomorphismsAll four theorems hold verbatim (only additive-subgroup and ideal-absorption properties are used)Confusing "subring" with "ideal" — a subring need not be an ideal
I⊈JI \not\subseteq J in Third Isom. / Correspondence Thm.Nothing guaranteedBijection/injectivity can fail, as shown in the counterexample
Non-commutative RRAll theorems hold with two-sided ideals throughoutOne-sided ideals do not generally support quotient ring structure
AA a subring but not an ideal, in Second Isom. Thm.Theorem still applies as stated (A+BA+B, A∩BA\cap B computed correctly)Assuming A⊴RA \trianglelefteq R is not needed — only BB needs to be an ideal

The mathematician behind it

Origins · 1 minEmmy Noether and how this was discovered
EN
Emmy Noether
1882–1935 · Göttingen, foundational abstract algebra

Emmy Noether transformed algebra from a subject about specific equations and formulas into the study of abstract structures — rings, ideals, modules — defined by their properties. The isomorphism theorems, though often first encountered for groups, were developed by Noether and her school in exactly the generality needed to unify ring theory, and her 1927 paper on the structure of ideal theory made the ideal-quotient correspondence a central organizing tool of the subject.

History
- 1921: Noether's paper "Idealtheorie in Ringbereichen" lays out ideal theory for commutative rings axiomatically.
- 1927: Noether formulates general isomorphism and correspondence results for rings and modules in a unified abstract framework.
- 1930: van der Waerden's Moderne Algebra, based on lectures by Noether and Artin, popularizes the isomorphism theorems in the now-standard textbook form.
- 1940s–50s: The theorems are absorbed into universal algebra and category theory as instances of general "first isomorphism theorem" statements for algebraic structures.
- Today: The theorems appear, essentially unchanged, as the opening structural results in every graduate and undergraduate ring theory course.
Do the isomorphism theorems require rings to be commutative?

No. All four theorems hold for arbitrary rings, commutative or not, as long as "ideal" means two-sided ideal. The proofs use only the additive group structure, the absorption property rI⊆IrI \subseteq I and Ir⊆IIr \subseteq I, and the universal property of quotients — none of these require commutativity. In fact these are the same theorems (suitably specialized) that hold for groups, replacing "normal subgroup" with "ideal." The only place commutativity becomes essential in ring theory is later, when discussing prime and maximal ideals via R/IR/I being an integral domain or field, since those notions depend on commutative cancellation properties.

Does the ring need to have a multiplicative identity for these theorems to hold?

No. The theorems as proved here use only additive-group reasoning plus the ideal absorption property and homomorphism properties φ(a+b)=φ(a)+φ(b)\varphi(a+b)=\varphi(a)+\varphi(b), φ(ab)=φ(a)φ(b)\varphi(ab)=\varphi(a)\varphi(b). None of the proofs invoke 1R1_R. If your convention requires ring homomorphisms to send 1R↦1S1_R \mapsto 1_S, then extra care is needed only in checking that induced maps between quotients respect identities, which they automatically do since φˉ(1R+K)=φ(1R)=1S\bar\varphi(1_R + K) = \varphi(1_R) = 1_S.

How does the First Isomorphism Theorem for rings relate to the one for groups?

A ring homomorphism is in particular a homomorphism of the underlying additive groups, so the group-theoretic First Isomorphism Theorem already gives an additive group isomorphism R/ker⁡φ≅im⁡φR/\ker\varphi \cong \operatorname{im}\varphi. The ring-theoretic content is the extra verification that this same bijection also respects multiplication, and that ker⁡φ\ker\varphi is not just a normal (here, every) additive subgroup but an ideal, so that R/ker⁡φR/\ker\varphi carries a ring structure at all. The proof given above rebuilds this from scratch rather than quoting the group version, to keep ring-multiplicativity checks explicit.

Can the Second Isomorphism Theorem fail if BB is only a subring, not an ideal?

Yes, dramatically. The proof needs B⊴RB \trianglelefteq R (or at least BB an ideal of A+BA+B) so that (A+B)/B(A+B)/B is a well-defined quotient ring and so that the kernel computation ker⁡φ=A∩B\ker\varphi = A \cap B goes through with φ(a)=a+B\varphi(a) = a+B actually being a ring homomorphism. If BB is merely a subring, A+BA+B need not even be closed under multiplication, so (A+B)/B(A+B)/B may not make sense as a quotient ring, and the entire construction collapses before it starts.

Is the bijection in the Correspondence Theorem also a ring isomorphism of some kind?

The Correspondence Theorem gives a bijection between two sets of ideals (a poset isomorphism, since it also preserves inclusions both ways), not directly a ring isomorphism. However, for each ideal JJ in the correspondence, the theorem additionally supplies a genuine ring isomorphism (R/I)/(J/I)≅R/J(R/I)/(J/I) \cong R/J, which is the Third Isomorphism Theorem attached to each matched pair. So the "lattice" correspondence is set/order-theoretic, while the quotient formula attached to it is ring-theoretic.

Why is it called the "Diamond" Isomorphism Theorem sometimes?

The name comes from the shape of the associated Hasse diagram: draw RR at top, AA and BB as two incomparable middle nodes, and A∩BA \cap B at the bottom, with A+BA+B sitting above both AA and BB. Extending upward to include A+BA+B above AA and BB, and A∩BA \cap B below both, produces a diamond-shaped lattice of four (or five, including RR) nodes, which is the visual mnemonic for the Second Isomorphism Theorem.

What goes wrong if I try to apply the Third Isomorphism Theorem with J⊆IJ \subseteq I instead of I⊆JI \subseteq J?

The statement doesn't parse: J/IJ/I is only defined when I⊆JI \subseteq J, since otherwise the cosets j+Ij + I for j∈Jj \in J need not partition consistently into a subset of R/IR/I in the intended sense (some "coset representatives" from JJ would not even be well-defined relative to II if I⊈JI \not\subseteq J, since JJ might not contain II at all, meaning J/IJ/I as a quotient makes no sense). You would instead want J⊆IJ \subseteq I paired with the different quotient (R/J)/(I/J)≅R/I(R/J)/(I/J) \cong R/I, which is the same theorem with the roles of II and JJ swapped.

Before this

Where this leads