Lagrange’s theorem is a fundamental result in group theory stating that, for a finite group and a subgroup , the number of elements of divides the number of elements of . More precisely,
where is the number of distinct left cosets of in , called the index of . The theorem follows by partitioning the group into equally sized cosets. It constrains possible subgroup sizes, determines basic properties of element orders, and underlies many counting arguments in finite group theory.
Statement and terminology
The order of a finite group is its number of elements, denoted . A subgroup is a subset that is itself a group under the operation inherited from .
For , the left coset of represented by is
The index counts the distinct left cosets. The number of right cosets is the same.
Lagrange’s theorem can therefore be stated in either of two equivalent forms:
- If is finite and , then .
- The order of a finite group equals the order of a subgroup multiplied by its index.
No commutativity assumption is required, and need not be a normal subgroup. Normality is needed to give the coset set a quotient-group operation, not to count its elements.
Proof by cosets
The standard proof uses two facts: every coset has as many elements as , and distinct cosets are disjoint.
For any , the map
is a bijection. It is surjective by the definition of , and injective because cancellation gives
Thus .
Next, suppose and intersect. There are such that
It follows that
which implies . Consequently, two left cosets are either equal or disjoint.
Every belongs to , since the identity element belongs to . Hence the distinct left cosets form a partition of . If there are of them, then
Equivalently, cosets are the equivalence classes of the relation
The theorem is therefore an instance of counting a finite set by equally sized equivalence classes.
Examples
An additive group of residues
In the additive group , a standard object in modular arithmetic, consider
Its distinct cosets are
There are four cosets, each with three elements, so
A nonnormal subgroup
The symmetric group , consisting of the permutations of three objects, has order . Its subgroup
has order , and hence index .
This subgroup is not normal: conjugation by a suitable permutation sends to another transposition. Nevertheless, its three left cosets partition into sets of size . This illustrates why normality is unnecessary for Lagrange’s theorem.
Consequences for element orders
The order of an element , written , is the smallest positive integer such that , when such an integer exists. The subgroup generated by ,
is a cyclic group and has exactly elements.
Applying Lagrange’s theorem gives
for every element of a finite group. In particular,
This restricts possible element orders but does not guarantee their existence. For example, a group of order can have elements only of orders or , but it need not have elements of all these orders.
Groups of prime order
If , where is a prime number, choose any . Its order divides and is greater than , so it must equal . Therefore
Thus every group of prime order is cyclic. In particular, every such group is an abelian group, and any two groups of the same prime order are isomorphic.
Applications to number theory
For a prime , the nonzero residue classes modulo form a multiplicative group of order . Therefore, if ,
This is the multiplicative form of Fermat’s little theorem.
More generally, the invertible residue classes modulo a positive integer form a group of order , where is Euler’s totient function. Lagrange’s theorem yields Euler’s theorem:
Why the converse fails
Lagrange’s theorem gives a necessary condition for a subgroup’s order, not a sufficient condition. If , there need not be a subgroup of order .
A standard counterexample is the alternating group , the group of even permutations of four objects. It has order but no subgroup of order .
To see this, suppose had order . Its index would be . Every subgroup of index is normal, so there would be a quotient homomorphism
onto a group of order .
Every -cycle must map to the identity: its image has cube equal to the identity, whereas the nonidentity element of a group of order does not. But the -cycles generate , so the homomorphism would be trivial, contradicting surjectivity.
There are important cases in which partial converses hold:
- Cyclic groups: A cyclic group of order has exactly one subgroup of order for every positive divisor of .
- Finite abelian groups: A finite abelian group has a subgroup of every order dividing its order, though that subgroup need not be unique.
- Prime divisors: Cauchy’s theorem guarantees an element, and therefore a cyclic subgroup, of order whenever the prime divides .
- Prime-power divisors: The Sylow theorems guarantee subgroups whose order is the largest power of a given prime dividing .
These existence statements require additional arguments; they do not follow merely by reversing Lagrange’s divisibility assertion.
Index, quotient groups, and group actions
Multiplicativity of index
For subgroups of a finite group,
This follows from
It can also be proved directly by choosing coset representatives, and the corresponding counting identity extends to infinite indices.
Quotient groups
If is normal in a finite group , its cosets form the quotient group . Lagrange’s formula becomes
Without normality, the cosets still form a set of this size, but multiplication of cosets need not be well defined.
The orbit–stabilizer theorem
Suppose acts on a set . For , its stabilizer is
The orbit of is in bijection with the left cosets of , through
Thus the orbit–stabilizer theorem gives, for finite ,
Lagrange’s theorem and orbit–stabilizer express the same underlying coset-counting principle. This connection makes the theorem useful in the study of group actions, symmetry, and permutation groups.
Infinite groups
For an infinite group, ordinary divisibility of finite orders is no longer the appropriate statement. The coset partition remains valid, however.
In the usual framework of set theory with the axiom of choice, choosing one representative from each left coset gives a bijection between and the Cartesian product of with the coset set. Consequently,
holds as an identity of cardinalities.
This identity has weaker numerical consequences than in the finite case because infinite cardinal multiplication does not permit ordinary cancellation. For example, the additive group and its subgroup are both countably infinite, but
Index therefore retains information that the cardinalities of the two groups alone do not determine.
Historical background
The theorem is named after Joseph-Louis Lagrange, whose work of 1770–1771 on algebraic equations examined how permutations of roots affect the values of expressions formed from them. His counting arguments anticipated the divisibility ideas later expressed in group-theoretic language.
Lagrange did not state the theorem in the modern language of arbitrary groups and subgroups. That formulation emerged with the nineteenth-century development of permutation groups, abstract algebra, and Galois theory.