First Isomorphism Theorem (for groups)

Field: Algebra · Level: First two years of university · Last updated: 2026-09-14

In one sentence: if φ\varphi is a structure-preserving map from a group GG to a group HH, then GG "divided by" the part that φ\varphi collapses (the kernel) has exactly the same structure as the part of HH that φ\varphi actually reaches (the image). G/ker⁡φ  ≅  im⁡φG / \ker\varphi \;\cong\; \im\varphi

In this chapter

What the theorem says

A map φ\varphi may send several elements of GG to the same place. The kernel ker⁡φ\ker\varphi is what decides which elements share a destination. The first isomorphism theorem says:

If you look at GG while treating elements that differ only by the kernel as "the same", what you see is indistinguishable, as a group, from the collection of destinations (the image).

In other words, the information that φ\varphi loses and the information that φ\varphi keeps separate cleanly.

This is the basic tool for understanding an abstractly defined quotient group G/NG/N as a concrete, familiar group.

Preparation: the words we need

Each of these notions gets its own chapter. Here we recall only what this chapter needs.

Definition (group)

A set GG with a "multiplication" (a binary operation a⋅ba \cdot b) is a group if (1) the operation is associative, (ab)c=a(bc)(ab)c = a(bc); (2) there is an identity ee with ea=ae=aea = ae = a; and (3) every aa has an inverse a−1a^{-1} with aa−1=a−1a=eaa^{-1} = a^{-1}a = e. Examples: the integers Z\Z under addition (identity 00, inverse of nn is −n-n); the nonzero reals R×\R^\times under multiplication.

Definition (homomorphism)

A map φ\varphi from a group GG to a group HH is a homomorphism if for all a,b∈Ga, b \in G φ(ab)=φ(a) φ(b).\varphi(ab) = \varphi(a)\,\varphi(b). "Multiply first and then map, or map first and then multiply — same answer." A homomorphism preserves the structure of the operation. It follows automatically that φ(e)=e\varphi(e) = e and φ(a−1)=φ(a)−1\varphi(a^{-1}) = \varphi(a)^{-1}.

Definition (kernel and image)

For a homomorphism φ:G→H\varphi : G \to H, ker⁡φ={ g∈G∣φ(g)=e },im⁡φ={ φ(g)∣g∈G }\ker\varphi = \set{\,g \in G \mid \varphi(g) = e\,}, \qquad \im\varphi = \set{\,\varphi(g) \mid g \in G\,} are the kernel and the image of φ\varphi. The kernel is the set of elements sent to the identity; the image is the set of values that actually occur. The kernel is a subgroup of GG, the image a subgroup of HH.

Definition (normal subgroup and quotient group)

A subgroup NN of GG is normal if gng−1∈Ng n g^{-1} \in N for every g∈Gg \in G and n∈Nn \in N. Given a normal subgroup NN, declare gg and g′g' to be "the same" when g−1g′∈Ng^{-1}g' \in N. The set of resulting bundles is written G/NG/N, and the bundle (coset) containing gg is written gNgN. When NN is normal, the rule (gN)(g′N)=(gg′)N(gN)(g'N) = (gg')N is a well-defined operation and makes G/NG/N a group, the quotient group.

Quotient groups look very abstract at first. The value of the first isomorphism theorem is that it translates the abstract G/NG/N into a group you already know.

Three examples first

Before stating the theorem, let us see the pattern "divide by the kernel, get the image" in three cases.

Example 1: the sign of a real number

From the multiplicative group R×\R^\times of nonzero reals to the group {+1,−1}\set{+1, -1} (under multiplication), define φ(x)={+1(x>0)−1(x<0).\varphi(x) = \begin{cases} +1 & (x > 0) \\ -1 & (x < 0). \end{cases} Since "positive times positive is positive, positive times negative is negative, negative times negative is positive", we have φ(xy)=φ(x)φ(y)\varphi(xy) = \varphi(x)\varphi(y): a homomorphism. The kernel is the set of xx with φ(x)=+1\varphi(x) = +1, i.e. the positive reals R>0\R_{>0}. The image is all of {+1,−1}\set{+1,-1}. The theorem says R×/R>0≅{+1,−1}\R^\times / \R_{>0} \cong \set{+1, -1}: "ignore magnitude (identify numbers that differ by a positive factor) and all that remains is the sign." Exactly what intuition suggests.

Example 2: wrapping the line around a circle

From the additive group R\R to the multiplicative group S1={z∈C∣∣z∣=1}S^1 = \set{z \in \C \mid |z| = 1} of complex numbers of absolute value 11, take φ(t)=e2πit.\varphi(t) = e^{2\pi i t}. The exponential law e2πi(s+t)=e2πise2πite^{2\pi i(s+t)} = e^{2\pi i s} e^{2\pi i t} says precisely that addition is sent to multiplication: a homomorphism. φ(t)=1\varphi(t) = 1 exactly when tt is an integer, so the kernel is Z\Z. The image is the whole circle S1S^1. The theorem: R/Z≅S1\R / \Z \cong S^1. "Roll up the real line, identifying points that differ by an integer, and you get a circle." It is the same idea as a clock face, where 13 o'clock and 1 o'clock are the same.

Example 3: the determinant

The invertible n×nn \times n real matrices GLn(R)GL_n(\R) form a group under matrix multiplication, and the determinant det⁡:GLn(R)→R×\det : GL_n(\R) \to \R^\times satisfies det⁡(AB)=det⁡Adet⁡B\det(AB) = \det A \det B, so it is a homomorphism. The kernel is the set of matrices with determinant 11, written SLn(R)SL_n(\R). The image is all of R×\R^\times (adjust one diagonal entry to get any nonzero value). The theorem: GLn(R)/SLn(R)≅R×GL_n(\R)/SL_n(\R) \cong \R^\times. "Ignore the determinant-one matrices, and what remains of an invertible matrix is a single number: its determinant."

In every example the kernel is "what φ\varphi cannot tell apart", the image is "what φ\varphi does tell apart", and dividing by the former leaves the latter.

Statement

Theorem (First Isomorphism Theorem)

Let φ:G→H\varphi : G \to H be a homomorphism of groups. Then

  1. ker⁡φ\ker\varphi is a normal subgroup of GG;
  2. the rule gN↦φ(g)gN \mapsto \varphi(g) (with N=ker⁡φN = \ker\varphi) defines a group isomorphism from the quotient G/ker⁡φG/\ker\varphi onto the image im⁡φ\im\varphi.

Hence G/ker⁡φ  ≅  im⁡φ.G / \ker\varphi \;\cong\; \im\varphi . In particular, if φ\varphi is surjective (its image is all of HH), then G/ker⁡φ≅HG/\ker\varphi \cong H.

"Isomorphic", written ≅\cong, means that a bijective homomorphism exists: after renaming elements, the two groups are literally the same.

Why it is true: the intuition

Think of φ\varphi as a light that projects the elements of GG onto HH. Some elements cast the same shadow: those g,g′g, g' with φ(g)=φ(g′)\varphi(g) = \varphi(g'). In that case φ(g−1g′)=φ(g)−1φ(g′)=e,\varphi(g^{-1}g') = \varphi(g)^{-1}\varphi(g') = e, so g−1g′g^{-1}g' lies in the kernel. Conversely, if g−1g′∈ker⁡φg^{-1}g' \in \ker\varphi then φ(g)=φ(g′)\varphi(g) = \varphi(g'). So:

"same shadow"   ⟺  \iff "differ by an element of the kernel"

This one line is the heart of the theorem. The quotient G/ker⁡φG/\ker\varphi is exactly "the world in which elements differing by the kernel are identified", so it matches the world of shadows, the image, one for one. And because φ\varphi preserves products, the match respects the group structure.

Proof

Write N=ker⁡φN = \ker\varphi. The proof has four steps; each begins by saying what it is going to show.

Step 1: the kernel is a normal subgroup

To show: if g∈Gg \in G and n∈Nn \in N then gng−1∈Ng n g^{-1} \in N.

Because φ\varphi is a homomorphism, φ(gng−1)=φ(g) φ(n) φ(g)−1=φ(g) e φ(g)−1=e.\varphi(g n g^{-1}) = \varphi(g)\,\varphi(n)\,\varphi(g)^{-1} = \varphi(g)\, e\, \varphi(g)^{-1} = e . So gng−1∈Ng n g^{-1} \in N, and the quotient group G/NG/N exists.

Step 2: the map ψ:G/N→H\psi : G/N \to H is well defined

To show: setting ψ(gN)=φ(g)\psi(gN) = \varphi(g) does not depend on the representative gg chosen for the coset.

Suppose gN=g′NgN = g'N. That means g−1g′∈Ng^{-1}g' \in N, so φ(g−1g′)=e\varphi(g^{-1}g') = e, i.e. φ(g)−1φ(g′)=e\varphi(g)^{-1}\varphi(g') = e, hence φ(g)=φ(g′)\varphi(g) = \varphi(g'). Changing the representative does not change the value.

Step 3: ψ\psi is a homomorphism

To show: ψ((gN)(g′N))=ψ(gN) ψ(g′N)\psi\big((gN)(g'N)\big) = \psi(gN)\,\psi(g'N).

By the definition of the quotient operation and the homomorphism property of φ\varphi, ψ((gN)(g′N))=ψ(gg′N)=φ(gg′)=φ(g)φ(g′)=ψ(gN) ψ(g′N).\psi\big((gN)(g'N)\big) = \psi(gg'N) = \varphi(gg') = \varphi(g)\varphi(g') = \psi(gN)\,\psi(g'N).

Step 4: ψ\psi is injective and its image is im⁡φ\im\varphi

Injective: if ψ(gN)=ψ(g′N)\psi(gN) = \psi(g'N) then φ(g)=φ(g′)\varphi(g) = \varphi(g'), so φ(g−1g′)=e\varphi(g^{-1}g') = e, so g−1g′∈Ng^{-1}g' \in N, i.e. gN=g′NgN = g'N. Distinct cosets go to distinct places.

Image: the values of ψ\psi are exactly the elements of the form φ(g)\varphi(g), so the image of ψ\psi is im⁡φ\im\varphi. Therefore ψ\psi, with its codomain restricted to im⁡φ\im\varphi, is a bijective homomorphism G/N→im⁡φG/N \to \im\varphi: an isomorphism.

Looking back, all we used was "φ\varphi preserves products" and the definition of the kernel. That this alone forces the abstract quotient to coincide with the image is what makes the theorem beautiful.

Where it is used

Corollary (counting)

If GG is finite, isomorphic groups have the same number of elements and G/NG/N has ∣G∣/∣N∣|G|/|N| elements, so ∣G∣=∣ker⁡φ∣⋅∣im⁡φ∣.|G| = |\ker\varphi| \cdot |\im\varphi| . "Size of GG = what was collapsed × what remains."

Common misconceptions

Misconception 1: “G/ker⁡φG/\ker\varphi is a subgroup of GG”

No. The elements of G/ker⁡φG/\ker\varphi are not elements of GG but bundles (cosets) of elements of GG. The theorem does not speak of a subgroup; it says that GG, regrouped, is isomorphic to the image.

Misconception 2: “≅\cong means ==”

Isomorphic means "same structure as a group", not equal as sets. In Example 1 the elements of R×/R>0\R^\times/\R_{>0} are the two sets "all positive reals" and "all negative reals", not the numbers +1+1 and −1-1 themselves.

Misconception 3: “the image is always all of HH”

The image is a subgroup of HH but need not be all of it. For φ:Z→Z\varphi : \Z \to \Z, n↦2nn \mapsto 2n, the image is the even integers. One may write G/ker⁡φ≅HG/\ker\varphi \cong H only when φ\varphi is surjective.

Misconception 4: “you can form a quotient by any subgroup”

G/NG/N is a group only when NN is normal. Kernels are always normal (Step 1), so there is no problem here; conversely every normal subgroup is the kernel of some homomorphism (the natural projection G→G/NG \to G/N has kernel NN).

Exercises

Exercise 1. Let φ:Z→Z\varphi : \Z \to \Z be φ(n)=2n\varphi(n) = 2n (additive groups). Check that φ\varphi is a homomorphism, find its kernel and image, and state what the theorem says.

Solution

φ(m+n)=2(m+n)=2m+2n=φ(m)+φ(n)\varphi(m+n) = 2(m+n) = 2m + 2n = \varphi(m) + \varphi(n), so φ\varphi is a homomorphism. 2n=02n = 0 only for n=0n = 0, so ker⁡φ={0}\ker\varphi = \set{0}. The image is the even integers 2Z2\Z. The theorem says Z/{0}≅2Z\Z/\set{0} \cong 2\Z: the integers and the even integers are isomorphic groups (via n↔2nn \leftrightarrow 2n), a phenomenon possible only for infinite sets.

Exercise 2. Let φ:Z/12Z→Z/4Z\varphi : \Z/12\Z \to \Z/4\Z send "the remainder mod 1212" to "its remainder mod 44" (re-reading a 12-hour clock with a 4-hour period). Explain why this is well defined, find the kernel and the image, and verify ∣G∣=∣ker⁡φ∣⋅∣im⁡φ∣|G| = |\ker\varphi|\cdot|\im\varphi|.

Solution

Since 44 divides 1212, numbers differing by a multiple of 1212 have the same remainder mod 44, so the map is well defined. The kernel consists of the remainders divisible by 44: {0,4,8}\set{0, 4, 8} (three elements). The image is all of {0,1,2,3}\set{0,1,2,3} (four elements). Indeed 12=3×412 = 3 \times 4. The theorem gives (Z/12Z)/{0,4,8}≅Z/4Z(\Z/12\Z)/\set{0,4,8} \cong \Z/4\Z.

Exercise 3. For n≥2n \ge 2, use the sign homomorphism sgn⁡:Sn→{+1,−1}\operatorname{sgn} : S_n \to \set{+1,-1} from the previous section to show that the number of even permutations is n!/2n!/2.

Solution

sgn⁡\operatorname{sgn} is a homomorphism (the sign of a composition is the product of the signs) with kernel AnA_n. For n≥2n \ge 2 a transposition exists and has sign −1-1, so the image is all of {+1,−1}\set{+1,-1}. The corollary gives n!=∣An∣⋅2n! = |A_n| \cdot 2, hence ∣An∣=n!/2|A_n| = n!/2.

References

Formal verification in Lean✓ verified 2026-09-14 · leanprover/lean4:v4.34.0-rc2

Below is this theorem written in the Lean 4 proof assistant with the Mathlib library. The computer checks every step of the proof mechanically.

lean/MathThemodel/FirstIsomorphism.lean

import Mathlib

/-!
# 準同型定理(群の第一同型定理) / First Isomorphism Theorem for groups

ページ: https://math.themodel.be/first-isomorphism-theorem/ja/ ,
https://math.themodel.be/first-isomorphism-theorem/en/

主張: 群の準同型 `φ : G →* H` に対して `G ⧸ φ.ker ≃* φ.range`。
ページ本文の証明(第1段〜第4段)に対応する補題を先に置き、最後に Mathlib の定理で同型を与える。
-/

namespace MathThemodel

variable {G H : Type*} [Group G] [Group H] (φ : G →* H)

/-- 補題1: 準同型の核 ker φ は G の正規部分群である。
    Lemma 1: the kernel of a homomorphism is a normal subgroup. -/
theorem ker_normal : φ.ker.Normal := inferInstance

/-- 補題2: 核 ker φ による剰余類の上で φ は矛盾なく定まり(well-defined)、
    誘導される写像 G / ker φ → H は [g] ↦ φ(g) で与えられる。
    Lemma 2: φ descends to the quotient, sending the coset [g] to φ(g). -/
theorem induced_map_apply (g : G) :
    QuotientGroup.kerLift φ (g : G ⧸ φ.ker) = φ g :=
  QuotientGroup.kerLift_mk φ g

/-- 補題3: 誘導された写像 G / ker φ → H は単射である。
    Lemma 3: the induced map is injective.
    証明: [a], [b] の像が等しいなら φ(a⁻¹ b) = φ(a)⁻¹ φ(b) = 1、
          すなわち a⁻¹ b ∈ ker φ、これは [a] = [b] を意味する。 -/
theorem induced_map_injective : Function.Injective (QuotientGroup.kerLift φ) := by
  intro a b h
  induction a using QuotientGroup.induction_on with
  | H a =>
    induction b using QuotientGroup.induction_on with
    | H b =>
      rw [QuotientGroup.kerLift_mk, QuotientGroup.kerLift_mk] at h
      rw [QuotientGroup.eq]
      exact φ.mem_ker.mpr (by rw [map_mul, map_inv, h, inv_mul_cancel])

/-- 準同型定理: G / ker φ ≅ im φ(群同型)。
    First Isomorphism Theorem: G / ker φ ≅ im φ. -/
noncomputable def firstIsomorphism : G ⧸ φ.ker ≃* φ.range :=
  QuotientGroup.quotientKerEquivRange φ

/-- 同型は [g] ↦ φ(g) で与えられる。
    The isomorphism is induced by φ itself. -/
theorem firstIsomorphism_apply (g : G) :
    ((firstIsomorphism φ) (g : G ⧸ φ.ker) : H) = φ g := rfl

/-- 系: φ が全射なら G / ker φ ≅ H。
    Corollary: if φ is surjective then G / ker φ ≅ H. -/
noncomputable def firstIsomorphismOfSurjective (hφ : Function.Surjective φ) :
    G ⧸ φ.ker ≃* H :=
  QuotientGroup.quotientKerEquivOfSurjective φ hφ

end MathThemodel