Skip to main content
Logo image

Section 14.2 The Class Equation

Let X be a finite G-set and XG be the set of fixed points in X; that is,
XG={xX:gx=x for all gG}.
Since the orbits of the action partition X,
|X|=|XG|+i=kn|Oxi|,
where xk,,xn are representatives from the distinct nontrivial orbits of X.
Now consider the special case in which G acts on itself by conjugation, (g,x)gxg1. The center of G,
Z(G)={x:xg=gx for all gG},
is the set of points that are fixed by conjugation. The nontrivial orbits of the action are called the conjugacy classes of G. If x1,,xk are representatives from each of the nontrivial conjugacy classes of G and |Ox1|=n1,,|Oxk|=nk, then
|G|=|Z(G)|+n1++nk.
The stabilizer subgroups of each of the xi’s, C(xi)={gG:gxi=xig}, are called the centralizer subgroups of the xi’s. From Theorem 14.11, we obtain the class equation:
|G|=|Z(G)|+[G:C(x1)]++[G:C(xk)].
One of the consequences of the class equation is that the order of each conjugacy class must divide the order of G.

Example 14.12.

It is easy to check that the conjugacy classes in S3 are the following:
{(1)},{(123),(132)},{(12),(13),(23)}.
The class equation is 6=1+2+3.

Example 14.13.

The center of D4 is {(1),(13)(24)}, and the conjugacy classes are
{(13),(24)},{(1432),(1234)},{(12)(34),(14)(23)}.
Thus, the class equation for D4 is 8=2+2+2+2.

Example 14.14.

For Sn it takes a bit of work to find the conjugacy classes. We begin with cycles. Suppose that σ=(a1,,ak) is a cycle and let τSn. By Theorem 6.16,
τστ1=(τ(a1),,τ(ak)).
Consequently, any two cycles of the same length are conjugate. Now let σ=σ1σ2σr be a cycle decomposition, where the length of each cycle σi is ri. Then σ is conjugate to every other τSn whose cycle decomposition has the same lengths.
The number of conjugate classes in Sn is the number of ways in which n can be partitioned into sums of positive integers. In the case of S3 for example, we can partition the integer 3 into the following three sums:
3=1+1+13=1+23=3;
therefore, there are three conjugacy classes. There are variations to problem of finding the number of such partitions for any positive integer n that are what computer scientists call NP-complete. This effectively means that the problem cannot be solved for a large n because the computations would be too time-consuming for even the largest computer.

Proof.

We apply the class equation
|G|=|Z(G)|+n1++nk.
Since each ni>1 and ni|G|, it follows that p must divide each ni. Also, p|G|; hence, p must divide |Z(G)|. Since the identity is always in the center of G, |Z(G)|1. Therefore, |Z(G)|p, and there exists some gZ(G) such that g1.

Proof.

By Theorem 14.15, |Z(G)|=p or p2. Suppose that |Z(G)|=p. Then Z(G) and G/Z(G) both have order p and must both be cyclic groups. Choosing a generator aZ(G) for G/Z(G), we can write any element gZ(G) in the quotient group as amZ(G) for some integer m; hence, g=amx for some x in the center of G. Similarly, if hZ(G)G/Z(G), there exists a y in Z(G) such that h=any for some integer n. Since x and y are in the center of G, they commute with all other elements of G; therefore,
gh=amxany=am+nxy=anyamx=hg,
and G must be abelian. Hence, |Z(G)|=p2.