Lagrange's Theorem Done Quick

2024-01-10 · 3 min read


  • done-quick
  • group-theory
  • math

Lagrange’s theorem is a relatively simple and powerful theorem. In its core, it says that the order of any subgroup divides the order of the original group. In fact, Lagrange also specifies the exact value of the quotient.

Theorem. (Lagrange) Given a group GG and a subgroup SGS \le G, denote with [G:S][G:S] the number of cosets of SS in GG. Then G=[G:S]S\left\lvert G \right\rvert = [G:S] \cdot \left\lvert S \right\rvert.

In order to prove this theorem, we need to do a bit of preparation.

The number of left and right cosets is the same

It is best to start by showing that the theorem is valid. If the number of right and left cosets is equal, then it is fine to emit the factor position.

In order to do so, we can construct a bijection between the set of left and right cosets. Let ff be a function from the right cosets to the left cosets, and define it as f(Sa)=a1Sf(Sa) = a^{-1} S.

It is simple prove that ff is a bijection. Consider the function gg from the left to right cosets, defined as g(aS)=Sa1g(aS) = S a^{-1}, and notice that gg is the inverse of ff:

(gf)(Sa)=g(f(Sa))=g(a1S)=S(a1)1=Sa(g \circ f) (Sa) = g(f(Sa)) = g(a^{-1} S) = S \left(a^{-1}\right)^{-1} = Sa

therefore ff is a bijection, hence the number of left and right cosets is equal.

Notice also that this proof works for infinite sets of cosets (simply, replace any mention of “number” by “cardinality”).


From this point, when referring to cosets we mean right cosets, although the statements are also true for left cosets.

Cosets are either disjoint or equal

Consider two cosets Sa,SbSa, Sb. We need to show that if SaSbSa \cap Sb \ne \varnothing, then Sa=SbSa = Sb. Let xSaSbx \in Sa \cap Sb.

Because xSax \in Sa, then x=s1ax = s_1 a for some s1Ss_1 \in S. Similarly, because xSbx \in Sb then x=s2bx = s_2 b for some s2Ss_2 \in S. Comparing both expressions yields:

s1a=s2bs1=s2ba1s21s1=ba1\begin{aligned} s_1 a &= s_2 b \\ s_1 &= s_2 b a^{-1} \\ {s_2}^{-1} s_1 &= b a^{-1} \end{aligned}

Since SS is a group, and s1,s2Ss_1, s_2 \in S, so is s21s1S{s_2}^{-1} s_1 \in S. Therefore, c=ba1Sc = b a^{-1} \in S. In order to show equality, we show double inclusion.

Let y=saSay = sa \in Sa. Notice that sc1b=sab1b=sa=ys c^{-1} b = s a b^{-1} b = sa = y. However, this also shows that ySby \in Sb, because y=(sc1)by = (sc^{-1}) b. Similarly, for y=sbSby = sb \in Sb, we have y=scay = sca, and therefore ySay \in Sa. This shows that if SaSbSa \cap Sb \ne \varnothing, then Sa=SbSa = Sb. In other words, cosets are either disjoint or equal.

Cosets partition GG

Notice that for any xGx \in G, xx appears in exactly one coset. Simply, this is the coset SxSx. Because cosets are either disjoint or equal, this means that there is no coset SxSx' for which xSxx \in Sx' and SxSxSx' \ne Sx.

Because each element in GG appears in exactly one coset, then the set of all cosets is a partition of GG.

Cosets have equal size

Consider some coset SaSa, and the original subgroup SS. The function mapping from SS to SaSa defined by f(s)=saf(s) = sa is a bijection. To show this, define the function gg mapping from SaSa to SS which is defined as g(sa)=sg(sa) = s. It is easy to see that gg is the inverse of ff:

(gf)(s)=g(sa)=s(g\circ f)(s) = g(sa) = s

so ff is a bijection. This shows that Sa=S\left\lvert Sa \right\rvert = \left\lvert S \right\rvert.

In particular, this is true for all cosets, and thus all cosets have the same size, of S\left\lvert S \right \rvert.

The Proof

Finally, we are ready to prove Lagrange’s theorem.

Given SGS \le G, denote with k=[G:S]k = [G:S] the number of cosets (this number is also referred as the index of SS). Because the cosets of SS partition GG, then G=Sa1Sa2SakG = S{a_1} \cup S{a_2} \cup \ldots \cup S{a_k} for some elements a1,,aka_1, \ldots, a_k. Moreover, because cosets are disjoint, then this is actually a disjoint union. Therefore, G=1ikSai\left\lvert G \right \rvert = \sum_{1 \le i \le k} \left\lvert S {a_i} \right \rvert. Finally, because Sai=S\left\lvert S {a_i} \right\rvert = \left\lvert S \right\rvert, we have:

G=1ikSai=1ikS=kS=[G:S]S \begin{aligned} \left\lvert G \right\rvert &= \sum_{1 \le i \le k} \left\lvert S {a_i} \right\rvert \\ &= \sum_{1 \le i \le k} \left\lvert S \right\rvert \\ &= k \cdot \left\lvert S \right\rvert = [G:S] \cdot \left\lvert S \right\rvert \end{aligned}

\blacksquare