What Is a Group?

Rotate a square by a quarter-turn and it looks exactly the same. Do it again — same square. Flip it across a diagonal — still the same square. The moves that leave the square looking unchanged can be chained together, undone, and combined in any order, and every combination is again one of the same moves. Hidden inside that simple observation is one of the most powerful ideas in all of mathematics: a group. A group is not a kind of number or a kind of shape — it is the abstract essence of symmetry and reversible combination, distilled down to four short rules.

You have already met groups without being told their name. The integers under addition form a group. The hours on a clock, where 11 + 3 = 2, form a group. The ways of shuffling a deck of cards form a group. The twists of a Rubik's cube form a group. Once you learn to spot the four rules, you start seeing the same skeleton everywhere — and anything you prove about "groups in general" instantly becomes true about all of these at once. That is the deal group theory offers: prove it once, use it everywhere.

In this page we pin down exactly what a group is — the four axioms of closure, associativity, identity, and inverse — see them satisfied in familiar systems, and (just as importantly) watch them fail in systems that look group-like but aren't. By the end you should be able to pick up any set with an operation and decide, rule by rule, whether it is a group.

A first example: adding on a clock

Take the numbers \{0, 1, 2, 3\} and add them the way a 4-hour clock would — wrapping around whenever you pass 4. This is modular arithmetic modulo 4, and the whole system is written \mathbb{Z}_4. So 2 + 3 = 5 \equiv 1, and 3 + 1 = 4 \equiv 0.

Watch how the four rules quietly hold. Add any two of these numbers and you land back inside the set \{0,1,2,3\} — never outside it (closure). Grouping the additions differently makes no difference (associativity). Adding 0 changes nothing (an identity). And every number has a partner that sends it back to 0: 1 + 3 = 0, 2 + 2 = 0 (an inverse). The whole thing is captured in a single grid — the Cayley table — where every row and every column is just the four numbers shuffled into a new order.

Notice two things in the table. The row and column headed by 0 simply copy the headers back unchanged — that is the identity at work. And no number ever repeats within a single row or column: each row is a permutation of \{0,1,2,3\}. That no-repeats property is closure and invertibility showing up visually, and it is the fingerprint of a group table.

The formal definition

A group is a set G together with a binary operation \ast (a rule combining two elements of G into one) satisfying all four of these axioms:

That is the entire definition — four lines, and every group in the universe obeys exactly these and nothing more. The economy is the point: because the rules are so few and so general, a theorem proved from them alone applies to numbers, symmetries, permutations, matrices, and cube twists all at once. Note that closure is sometimes folded silently into the phrase "binary operation on G," but it is worth stating out loud, because it is the axiom most often violated by accident.

Worked examples: is it a group?

The only way to build intuition is to test the axioms one at a time. Let's run three genuine groups past the checklist.

The integers under addition, (\mathbb{Z}, +). Closure: the sum of two integers is an integer. Associativity: ordinary addition is associative. Identity: 0, since a + 0 = a. Inverse: the negative -a, since a + (-a) = 0. All four hold — (\mathbb{Z}, +) is a group.

The integers mod n under addition, (\mathbb{Z}_n, +). Same story on a clock face. Closure by wrapping around; associativity inherited from ordinary addition; identity 0; and the inverse of a is n - a (with 0 its own inverse), since a + (n - a) = n \equiv 0. A group, and a finite one — it has exactly n elements.

The nonzero rationals under multiplication, (\mathbb{Q}^{\times}, \times). Closure: a product of nonzero fractions is a nonzero fraction. Associativity: multiplication is associative. Identity: 1. Inverse: the reciprocal 1/a, which exists precisely because we threw 0 out — 0 has no reciprocal, so it would have wrecked the inverse axiom had we kept it. A group, and the reason we insist on "nonzero."

Non-examples: when the rules break

Understanding a definition means knowing what it excludes. Each of these almost makes it, and fails on exactly one axiom.

The natural numbers under addition, (\mathbb{N}, +). Closure holds, associativity holds, and (if we include 0) there is an identity. But inverses fail: there is no natural number you can add to 3 to get back to 0, because -3 is not a natural number. So close — and yet not a group. This single missing axiom is exactly why we enlarge \mathbb{N} to \mathbb{Z}.

The integers under subtraction, (\mathbb{Z}, -). Here the casualty is associativity. Compare (8 - 5) - 2 = 3 - 2 = 1 with 8 - (5 - 2) = 8 - 3 = 5. Different answers, so the operation is not associative and this is not a group — even though closure holds and it looks perfectly reasonable at a glance.

The integers under multiplication, (\mathbb{Z}, \times). Closure, associativity, and identity (1) all hold, but inverses fail again: the reciprocal of 2 is \tfrac12, which is not an integer. Only 1 and -1 have integer inverses, so the set as a whole is not a group under multiplication.

Abelian and non-abelian groups

Notice what is not in the list of axioms: commutativity. Nothing requires a \ast b = b \ast a. When it happens to hold anyway — as it does for every example above — we call the group abelian, after Niels Henrik Abel. When it can fail, the group is non-abelian, and these are where group theory gets its richest structure.

The smallest non-abelian example is the symmetries of a triangle (the group D_3): rotating and then flipping generally lands you somewhere different from flipping and then rotating. Order of operations matters — exactly as it does for a Rubik's cube, where "twist top, then twist front" scrambles the cube differently from "twist front, then twist top." Addition on a clock is abelian; the physical act of manipulating an object in space usually is not.

The two axioms learners most often fumble are identity and inverse — and the trap is the word every. It is not enough for some elements to have inverses; each and every element must have one. The integers under multiplication feel almost like a group precisely because 1 and -1 do have inverses — but 2 does not, and a single element without an inverse is enough to disqualify the whole set.

Equally, the identity must be two-sided: the definition demands e \ast a = a and a \ast e = a, and the inverse must satisfy a \ast a^{-1} = e and a^{-1} \ast a = e. In an abelian group this is automatic, but in a non-abelian group you genuinely have to check both sides. And do not silently skip associativity: subtraction shows how an operation can look wholesome and still fail it. When someone hands you a set and an operation, march through all four axioms deliberately — one overlooked rule is the difference between a group and a near-miss.

The word "group" (groupe) was coined by Évariste Galois (1811–1832), a French teenager with a temper and a cause. Trying to settle a 300-year-old puzzle — why there is no formula (like the quadratic formula) to solve the general fifth-degree equation by radicals — Galois realised the answer lay not in the equation's roots but in the symmetries among them. Those symmetries formed a structure he was the first to name and study: the group.

He was expelled, jailed twice for his republican politics, and killed in a duel at the age of twenty. The night before, convinced he would die, he scrawled his mathematical testament in a frantic letter — margins crammed with "I have no time, I have no time." It took the world decades to understand what he had written. Today his idea underpins everything from the classification of finite simple groups to the security of the internet — not bad for a few pages dashed off by candlelight before dawn. The Rubik's cube in your drawer is, quite literally, a physical model of a Galois group.