Skip to content

Groups and the Group Axioms

In this lesson you’ll learn the definition of a group, how to verify it, and the first few theorems that follow from the axioms alone. This is the most important definition in the section, and everything after it is either an example or a consequence.

A group is a set GG together with a binary operation * satisfying four conditions.

Written out formally, (G,)(G, *) is a group when:

  1. Closure. For all a,bGa, b \in G, abGa * b \in G.
  2. Associativity. For all a,b,cGa, b, c \in G, (ab)c=a(bc)(a * b) * c = a * (b * c).
  3. Identity. There exists eGe \in G with ea=ae=ae * a = a * e = a for all aGa \in G.
  4. Inverses. For each aGa \in G there exists a1Ga^{-1} \in G with aa1=a1a=ea * a^{-1} = a^{-1} * a = e.

That’s the whole definition. Four lines, and a century of mathematics comes out of it.

Two remarks on how to read it. Closure is often folded into the phrase “binary operation on GG,” so some books list only three axioms. And commutativity is absent on purpose: a group where ab=baa * b = b * a always holds is called abelian, and leaving it optional is what allows groups to describe physical symmetry, where order genuinely matters.

The order of GG, written G|G|, is the number of elements. Z6=6|\mathbb{Z}_6| = 6. The integers under addition form an infinite group.

Be careful: “order” also refers to a property of a single element, defined in a later lesson. Context tells them apart, but the collision is real and worth flagging now.

Two styles appear throughout, and mixing them up causes genuine errors.

  • Multiplicative. Write abab for the operation, ee or 11 for the identity, a1a^{-1} for inverses, ana^n for repetition. Used for general groups.
  • Additive. Write a+ba + b, 00 for the identity, a-a for inverses, nana for repetition. Used when the group is abelian and the operation genuinely is addition.

The trap: in the additive group Z\mathbb{Z}, the inverse of 55 is 5-5, not 15\frac{1}{5}. The symbol a1a^{-1} means “whatever undoes aa,” and what that is depends entirely on the operation.

The value of an abstract definition is that you can prove things from it directly. Here are three, and notice that each proof uses only the four axioms.

The identity is unique.

Suppose ee and ff are both identities. Then

e=ef=fe = e * f = f

The first equality holds because ff is an identity, the second because ee is. So e=fe = f. There is only ever one identity, which justifies calling it the identity.

Inverses are unique.

Suppose bb and cc are both inverses of aa. Then

b=be=b(ac)=(ba)c=ec=cb = b * e = b * (a * c) = (b * a) * c = e * c = c

Every step is one axiom: identity, definition of cc, associativity, definition of bb, identity. This is why a1a^{-1} is well-defined notation.

Cancellation.

If ab=aca * b = a * c then b=cb = c.

Operate on the left by a1a^{-1}:

a1(ab)=a1(ac)    (a1a)b=(a1a)c    eb=ec    b=ca^{-1} * (a * b) = a^{-1} * (a * c) \implies (a^{-1} * a) * b = (a^{-1} * a) * c \implies e * b = e * c \implies b = c

Cancellation is exactly what inverses buy you. It is also why every row of a Cayley table contains each element exactly once: if a row repeated an element, cancellation would force two column headings to be equal.

The shoe-sock rule.

(ab)1=b1a1(ab)^{-1} = b^{-1}a^{-1}

The order reverses. To undo “put on socks, then shoes,” you take off shoes, then socks. Check it:

(ab)(b1a1)=a(bb1)a1=aea1=aa1=e(ab)(b^{-1}a^{-1}) = a(bb^{-1})a^{-1} = aea^{-1} = aa^{-1} = e

In an abelian group the reversal is invisible, which is why this is a favourite exam question.

Also (a1)1=a(a^{-1})^{-1} = a: undoing an undo returns you to the start.

A reliable order of attack:

  1. Name the set and the operation. Both. “Is the integers a group?” is not a question.
  2. Closure. Try to break it with a specific pair.
  3. Associativity. Usually inherited from a known operation. If the operation is novel, expand both groupings.
  4. Identity. Solve ae=aa * e = a for ee, then verify the other side.
  5. Inverses. Solve ab=ea * b = e for bb, then check bb is really in the set.

Step 5 is where most candidates die. The formula for the inverse exists, but the answer lands outside the set: 13\frac{1}{3} is a perfectly good number and not an integer.

Example 1: Is (Z,+)(\mathbb{Z}, +) a group?

Solution.

  • Closure. Integer plus integer is an integer. ✓
  • Associativity. Standard property of addition. ✓
  • Identity. 00, since 0+a=a+0=a0 + a = a + 0 = a. ✓
  • Inverses. For aa, take a-a, which is an integer, and a+(a)=0a + (-a) = 0. ✓

Yes, and it is abelian since a+b=b+aa + b = b + a. It is infinite.

Example 2: Is (Q,×)(\mathbb{Q}, \times) a group?

Solution. Closure ✓, associativity ✓, identity 11 ✓. Inverses: for a0a \ne 0 the inverse is 1a\frac{1}{a}, a rational. But 00 has no inverse, since 0b=010 \cdot b = 0 \ne 1 for every bb.

No. One element without an inverse is fatal.

Remove it and (Q{0},×)(\mathbb{Q} \setminus \{0\}, \times) is a group. Note you must check closure again after removing an element: a product of nonzero rationals is nonzero, so it holds.

Example 3: Is (Z5{0},×mod5)(\mathbb{Z}_5 \setminus \{0\}, \times \bmod 5) a group?

Solution. The set is {1,2,3,4}\{1, 2, 3, 4\}.

Closure: compute a few. 23=612 \cdot 3 = 6 \equiv 1, 24=832 \cdot 4 = 8 \equiv 3, 34=1223 \cdot 4 = 12 \equiv 2, 44=1614 \cdot 4 = 16 \equiv 1. Nothing ever gives 00, because 55 is prime and a product of two numbers not divisible by 5 is not divisible by 5. ✓

Associativity ✓ (inherited from integer multiplication). Identity 11 ✓.

Inverses: 11=11^{-1} = 1, 21=32^{-1} = 3 (since 23=612 \cdot 3 = 6 \equiv 1), 31=23^{-1} = 2, 41=44^{-1} = 4 (since 16116 \equiv 1). ✓

Yes, a group of order 4.

Now contrast Z6{0}={1,2,3,4,5}\mathbb{Z}_6 \setminus \{0\} = \{1,2,3,4,5\}: here 23=602 \cdot 3 = 6 \equiv 0, so closure fails immediately. The primality of the modulus is what makes the difference, and that observation grows into the theory of fields later in the section.

Example 4: Is the set of 2×22\times 2 real matrices a group under multiplication?

Solution. Closure ✓, associativity ✓, identity I=(1001)I = \begin{pmatrix} 1 & 0 \\ 0 & 1\end{pmatrix} ✓.

Inverses fail. The matrix (1111)\begin{pmatrix} 1 & 1 \\ 1 & 1 \end{pmatrix} has determinant 00 and no inverse.

No. Restricting to matrices with nonzero determinant gives GL2(R)GL_2(\mathbb{R}), the general linear group, which is a group: det(AB)=det(A)det(B)0\det(AB) = \det(A)\det(B) \ne 0 gives closure, and invertible matrices have inverses by definition. It is non-abelian, since

(1101)(1011)=(2111),(1011)(1101)=(1112)\begin{pmatrix} 1 & 1 \\ 0 & 1\end{pmatrix}\begin{pmatrix} 1 & 0 \\ 1 & 1\end{pmatrix} = \begin{pmatrix} 2 & 1 \\ 1 & 1\end{pmatrix}, \qquad \begin{pmatrix} 1 & 0 \\ 1 & 1\end{pmatrix}\begin{pmatrix} 1 & 1 \\ 0 & 1\end{pmatrix} = \begin{pmatrix} 1 & 1 \\ 1 & 2\end{pmatrix}

Example 5: Prove that in any group, (a1)1=a(a^{-1})^{-1} = a.

Solution. By definition, (a1)1(a^{-1})^{-1} is the unique element xx with a1x=xa1=ea^{-1} x = x a^{-1} = e.

But aa itself satisfies this: a1a=ea^{-1} a = e and aa1=ea a^{-1} = e, straight from the definition of a1a^{-1}.

Since inverses are unique (proved above), (a1)1=a(a^{-1})^{-1} = a. ∎

Notice the shape of the argument: to identify an inverse, exhibit something that works and cite uniqueness. That pattern recurs constantly.

Example 6: A group with exactly one element.

Solution. Take G={e}G = \{e\} with ee=ee * e = e. Closure ✓, associativity ✓ (only one triple to check), identity ee ✓, inverse of ee is ee ✓.

This is the trivial group. It’s not interesting, but it is a group, and it matters for the same reason the empty set matters: theorems have to be true in the degenerate case too. There is no group with zero elements, since axiom 3 demands an identity.

Example 7: Where associativity fails and nothing else does.

Let G={1,2,3}G = \{1, 2, 3\} with a\*b=a \* b = the larger of the two.

Solution. Closure ✓. Identity: 1a=a1 * a = a for all aa, so 11 works ✓. Associativity: taking the maximum is associative ✓.

Inverses: we need ab=1a * b = 1, but the maximum of 22 and anything is at least 22. So 22 has no inverse. Not a group.

This structure, closed and associative with an identity but without inverses, is called a monoid. String concatenation is the standard example: you can join strings and the empty string does nothing, but you cannot un-join.

Rubik’s cube. The moves form a group of order 43,252,003,274,489,856,000. Closure: any two sequences compose into a sequence. Identity: doing nothing. Inverses: run the sequence backwards. Associativity: concatenation of move lists. Solvers exploit commutators aba1b1aba^{-1}b^{-1}, which are guaranteed to be the identity if aa and bb commute, and which in practice disturb only a small region of the cube.

Cryptography. Diffie-Hellman key exchange runs inside (Zp{0},×)(\mathbb{Z}_p \setminus \{0\}, \times), exactly the group from Example 3 with a large prime. The whole protocol depends on that set being a group, and its security on the difficulty of reversing exponentiation in it.

Computer graphics and robotics. Rotations in 3D form the group SO(3)SO(3), which is non-abelian: turn a die about xx then zz, versus zz then xx, and the results differ. Because rotation composition is group multiplication, quaternions are preferred over stacked Euler angles, since they compose without the degeneracies that plague angle triples.

Physics. Noether’s theorem ties each continuous symmetry group of a system to a conserved quantity. The Standard Model is specified by naming a group, SU(3)×SU(2)×U(1)SU(3) \times SU(2) \times U(1); the particles are essentially bookkeeping for how that group acts.

Music. The twelve pitch classes under transposition form Z12\mathbb{Z}_{12}. Adding 7 repeatedly cycles through all twelve notes, which is the circle of fifths, and the reason it works is that 7 and 12 share no common factor. That fact reappears as a theorem two lessons from now.

Which of these is NOT one of the four group axioms?
Why is (ℚ, ×) not a group?
In a group, what does (ab)⁻¹ equal?
In the group (ℤ, +), what is the inverse of 7?
A structure is closed and associative and has an identity, but some element lacks an inverse. What is it?
Why does the set {1, 2, 3, 4} form a group under multiplication mod 5, while {1, 2, 3, 4, 5} fails under multiplication mod 6?