Pólya Enumeration Theorem
Catalog
Introduction
In high school combinatorics, we often encounter problems that require us to count the distinct colorings of an object with certain symmetrical properties, for example, a magiccube. How do we solve the problem?
The Burnside’s lemma might help us with this challenge. Consider all the possible transformations that the cube can undergo, for example, rotate it along the axis of two opposite faces by 9 0 ∘ 90^\circ 90∘, and the cube is of course considered equivalent to the original cube after these transformations. For each of these “transformations”, we count the number of colorings such that the cube still looks identical after the transformation when we view it from the same angle, and the average of these values, according to the Burnside’s lemma, turns out to be the number of distinct colourings of the cube.
Preliminary knowledge
Permutation Group
To understand Burnside’s lemma more thoroughly, we need some concepts from group theory to frame the problem more formally.
The “transformations” that we just mentioned are more formally known as permutations, which can be seen as a bijection from a finite set to itself. Consider the following necklace with 6 beads
a
1
−
a
2
−
a
3
a
6
−
a
1
−
a
2
∣
∣
=
>
∣
∣
a
6
−
a
5
−
a
4
a
5
−
a
4
−
a
3
\begin{aligned} a_1-a_2-a_3&\space\space\space\space\space\space\space\space a_6-a_1-a_2\\ |\space\space\space\space\space\space\space\space\space\space\space\space\space\space\space\space\space\space|&=>|\space\space\space\space\space\space\space\space\space\space\space\space\space\space\space\space\space\space|\\ a_6-a_5-a_4&\space\space\space\space\space\space\space\space a_5-a_4-a_3 \end{aligned}
a1−a2−a3∣ ∣a6−a5−a4 a6−a1−a2=>∣ ∣ a5−a4−a3
Here we rotate the necklace clockwisely by 1 beads, and we can describe this permutation p as
f
p
=
(
a
1
,
a
2
,
…
,
a
6
a
6
,
a
1
,
…
,
a
5
)
f_p=\begin{pmatrix}a_1,a_2,\dots,a_6\\\ a_6,a_1,\dots,a_5 \end{pmatrix}
fp=(a1,a2,…,a6 a6,a1,…,a5)
We define the multiplication of two permutations as carrying out the two permutations on the set in a consecutive manner. More formally, if we have two permutations p1 and p2 acting on set
S
=
{
a
1
,
a
2
,
…
,
a
n
}
S=\lbrace a_1,a_2,\dots,an\rbrace
S={a1,a2,…,an}, where
f
p
1
=
(
a
1
,
a
2
,
…
,
a
n
a
p
1
(
1
)
,
a
p
1
(
2
)
,
…
,
a
p
1
(
n
)
)
,
f
p
2
=
(
a
p
1
(
1
)
,
a
p
1
(
2
)
,
…
,
a
p
1
(
n
)
a
p
2
(
1
)
,
a
p
2
(
2
)
,
…
,
a
p
2
(
n
)
)
t
h
e
n
f
p
1
∘
f
p
2
=
(
a
1
,
a
2
,
…
,
a
n
a
p
2
(
1
)
,
a
p
2
(
2
)
,
…
,
a
p
2
(
n
)
)
f_{p_1}=\begin{pmatrix}a_1,a_2,\dots,a_n\\\ a_{p_1(1)},a_{p_1(2)},\dots,a_{p_1(n)} \end{pmatrix}, f_{p_2}=\begin{pmatrix}a_{p_1(1)},a_{p_1(2)},\dots,a_{p_1(n)}\\\ a_{p_2(1)},a_{p_2(2)},\dots,a_{p_2(n)} \end{pmatrix}\\ then\space f_{p_1}\circ f_{p_2}=\begin{pmatrix}a_1,a_2,\dots,a_n\\\ a_{p_2(1)},a_{p_2(2)},\dots,a_{p_2(n)} \end{pmatrix}
fp1=(a1,a2,…,an ap1(1),ap1(2),…,ap1(n)),fp2=(ap1(1),ap1(2),…,ap1(n) ap2(1),ap2(2),…,ap2(n))then fp1∘fp2=(a1,a2,…,an ap2(1),ap2(2),…,ap2(n))
This multiplication is associative: no matter in what order you rotate the necklace by 1 beads, 2 beads and 3 beads clockwisely, it is always equivalent to rotating the necklace by 6 beads clockwisely.
The multiplication is closed: since every permutation represents a bijection between a set and itself, it always map back to itself no matter how many times we perform the permutations.
There is an identity element, namely the permutation
f
I
=
(
a
1
,
a
2
,
…
,
a
n
a
1
,
a
2
,
…
,
a
n
)
f_I=\begin{pmatrix}a_1,a_2,\dots,a_n\\\ a_1,a_2,\dots,a_n \end{pmatrix}
fI=(a1,a2,…,an a1,a2,…,an)
such that for any permutation
p
p
p,
p
∘
I
=
I
∘
p
=
p
p\circ I=I\circ p=p
p∘I=I∘p=p.
There is an inverse of each permutation: we simply swap the 2 rows.
Thus, these permutations form a group, known as the permutation group.
Fixed Points
For each of these permutations,the colorings that the cube still looks identical after the transformation when we view it from the same angle are known as Fixed Points, since they are essentially the elements that map back to themselves if we consider the permutation group as a group of functions.
Again, let us go back to the example of the necklace consists of 6 beads. Suppose there are 2 beads of type x , y , a n d z x,y,and\space z x,y,and z each. Consider the permutation represented by rotating the necklace 3 bead clockwisely, how many diffent ways can we arrange the beads such that the ordered set { a 1 , a 2 , … , a 6 } \lbrace a_1,a_2,\dots,a_6\rbrace {a1,a2,…,a6} map back to itself after the permutation?
Apparently, we have a 1 = a 4 , a 2 = a 5 , a 3 = a 6 a_1=a_4,a_2=a_5,a_3=a_6 a1=a4,a2=a5,a3=a6 in this case, which means that { a 1 , a 4 } \lbrace a_1,a_4\rbrace {a1,a4}, { a 2 , a 5 } \lbrace a_2,a_5\rbrace {a2,a5}, { a 3 , a 6 } \lbrace a_3,a_6\rbrace {a3,a6} are of the same type respectively, thus there are 3!=6 fixed points, since we can assign the type to the 3 pairs arbitrarily.
Lagrange’s Theorem in Group Theory
Lagrange’s theorem in group theory states that for any finite group G G G, the order (number of elements) of every subgroup of G G G divides the order of G G G.
Firstly, for any group G G G of order n n n, we have 2 trivial subgroups: itself and the group that only contains the identity element, and they clearly do satisfy the statement since n ∣ n n|n n∣n and 1 ∣ n 1|n 1∣n.
Now, suppose we have a non-trivial subgroup H H H of G G G, we choose an element g 1 ∈ G g_1\in G g1∈G and g 1 ∉ H g_1\notin H g1∈/H. Consider the set generated by multipying each element of H H H by g 1 g_1 g1(with g 1 g_1 g1 being the left factor). Such sets are more formally known as the left cosets of H H H in G G G. We have a similar definition for right cosets as well. Note that H H H is both a left coset and a right coset of itself in G G G.
We claim that
H
H
H and
g
1
H
g_1H
g1H has no common element. Suppose there is a common element
H
j
H_j
Hj, then we have:
g
1
∗
H
i
=
H
j
f
o
r
s
o
m
e
H
i
∈
H
g
1
∗
H
i
∗
(
H
i
)
−
1
=
H
j
∗
(
H
i
)
−
1
g
1
∗
I
=
H
k
f
o
r
s
o
m
e
H
k
∈
H
g
1
=
H
k
f
o
r
s
o
m
e
H
k
∈
H
\begin{aligned} g_1*H_i&=H_j\space for\space some\space H_i\in H\\ g_1*H_i*(H_i)^{-1}&=H_j*(H_i)^{-1}\\ g_1*I&=H_k\space for\space some\space H_k\in H\\ g_1&=H_k\space for\space some\space H_k\in H\\ \end{aligned}
g1∗Hig1∗Hi∗(Hi)−1g1∗Ig1=Hj for some Hi∈H=Hj∗(Hi)−1=Hk for some Hk∈H=Hk for some Hk∈H
The third line follows since
H
H
H is a subgroup(so it is closed), and this proves contradictory since we have chosen
g
1
∉
H
g_1\notin H
g1∈/H.
We can also show that the set
g
1
H
g_1H
g1H has
∣
H
∣
|H|
∣H∣ elements by contradiction. Suppose there are 2 distinct elements
H
i
,
H
j
∈
H
H_i,H_j\in H
Hi,Hj∈H and
g
1
∗
H
i
=
g
1
∗
H
j
g_1*H_i=g_1*H_j
g1∗Hi=g1∗Hj.
(
g
1
)
−
1
∗
g
1
∗
H
i
=
(
g
1
)
−
1
∗
g
1
∗
H
j
I
∗
H
i
=
I
∗
H
j
H
i
=
H
j
\begin{aligned} (g_1)^{-1}*g_1*H_i&=(g_1)^{-1}*g_1*H_j\\ I*H_i&=I*H_j\\ H_i&=H_j \end{aligned}
(g1)−1∗g1∗HiI∗HiHi=(g1)−1∗g1∗Hj=I∗Hj=Hj
This again proves contradictory. Thus, the order of the left coset
g
1
H
g_1H
g1H is equal to the order of the subgroup
H
H
H.
We can then choose an element g 2 ∈ G g_2\in G g2∈G and g 2 ∉ H ∪ g 1 H g_2\notin H\cup g_1H g2∈/H∪g1H, and by a similar argument, we can prove that g 2 H g_2H g2H has no common element with H H H or g 1 H g_1H g1H and is of order ∣ H ∣ |H| ∣H∣. Repeat this process until we cannot find an element that lies outside of all these subsets, and we would have partitioned the group H H H into an arbitrary number of sets of order ∣ H ∣ |H| ∣H∣. This concludes the proof for Lagrange Theorem.
We will continue onto the proof for the Orbit-Stabilizer Theorem and how it leads to Burnside’s lemma and consequently Pólya Enumeration Theorem in the second half.