Reflection Groups

Arun Ram
Department of Mathematics and Statistics
University of Melbourne
Parkville, VIC 3010 Australia
aram@unimelb.edu.au

Last update: 20 March 2012

The groups G(r,p,n)

The dihedral group G(r,r,2) of order 2r

The symmetric group

The symmetric group is the group of permutations of 1,2,...,n, i.e. the group of bijections w: {1,2,...,n} {1,2,...,n} under the operation of composition. There are multiple notations for permutations:

  1. Two line notation: where w(i) in the second line is below i in the first line. w= 1 2 3 4 5 6 4 1 6 2 5 3
  2. Cycle notation: Where (i1,i2,...,ir) indicates w(i1) = i2,   w(i2) = i3,   ...,   w(ir-1) = ir,   w(ir) = i1. Cycles of length 1 are usually dropped from the notation. w=(142)(36)(5) =(142)(36).
  3. Matrix notation: where the w(i)th entry of the ith column is 1 and all other entries are 0. w= 0 1 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 1 0 0 0 0 0 0 0 0 0 1 0 0 0 1 0 0 0
  4. Diagram notation: where the ith dot in the top row is connected to the w(i)th dot in the bottom row. w=
The symmetric group Sn is the group of n×n permutation matrices and Card(Sn) = n!.

The transpositions in Sn are the permutations (i,j),   1i<jn (in cycle notation). The simple transpositions are si=(i,i+1) = i i+1   , 1in-1.

The symmetric group Sn can be presented by generators s1,s2,...,sn-1 and relations sisj = sjsi, if   |i-j|>1, sisi+1si = si+1sisi+1, 1in-2, si2 = 1, 1in-1.

Proof.
There are three things to show:
  1. The simple transpositions in Sn satisfy the given relations.
  2. The simple transpositions generate Sn.
  3. The group given by generators s1,...,sn-1 and relations as in the statement of the theorem has order n!.
  1. The following pictures show that the simple transpositions si=(i,i+1),   1in-2 satisfy the given relations. = = = = =
  2. The diagram of any permutation can be "stretched" to guarantee that no more than two edges cross at any given point. Then w can be written as a product of simple transpositions corresponding to these crossings. w= = = s1s3 s5s4 s5s3.
  3. Let Gn be the free group generated by symbols s1,s2,...,sn-1 modulo the relations in the statement of the theorem. Let Gn-1 be the subgroup generated by s1,s2,...,sn-2. We will show that
    1. Every element wGn can be written in the form w= w1sn-1w2, with w1,w2 Gn-1 .
    2. Every element wGn can be written in the form w= aan-1 sn-2sk,   1kn, and aGn-1.

    1. Since sn-12=1 every element wGn can be written in the form w= w1sn-1 w2sn-1 w3sn-1 wl-1 sn-1 wl, where   wjGn-1. By induction, either w2Gn-2 or w2=asn-2b where a,bGn-2. In the first case w = w1sn-1 w2sn-1 w3sn-1 wl-1 sn-1 wl = w1w2 sn-1 sn-1 w3 sn-1 wl-1 sn-1 wl = w1w2 sn-1 w3 sn-1 wl-1 sn-1 wl, and in the second case w = w1 sn-1 w2 sn-1 w3 sn-1 wl-1 sn-1 wl = w1 sn-1a sn-2b sn-1 w3 sn-1 wl-1 sn-1 wl = w1a sn-1 sn-2 sn-1b w3 sn-1 wl-1 sn-1 wl = w1a sn-2 sn-1 sn-2b w3 sn-1 wl-1 sn-1 wl = w1 sn-1 w3 sn-1 wl-1 sn-1 wl where w1 = w1a sn-2 Gn-1 and w3 = sn-2b w3 Gn-1 . In either case the number of sn-1 factors in the expression of w has been reduced. The statement in (1) follows by induction.
    2. By (1) any element wGn is either an element of Gn-1 or can be written in the form w=w1 sn-1 w2, with w1,w2 Gn-1 . In the first case w=w sn-1 sl with l=n and the statement is satisfied. If w= w1 sn-1 w2 then, by induction, w2=a sn-2 sn-3 sl, for some 1ln-2 and aGn-2. So w=w1 sn-1a sn-2 sl = w1a sn-1 sl = w1 sn-1 sl, where w1 = w1a Gn-1 . Statement (2) follows.
    From (2) it follows that Card(Gn) Card(Gn-1)n, and, by induction, that Card(Gn) n!. Since Sn is a quotient of Gn and Card(Sn) = n! we have GnSn.

A partition of n is a collection of boxes in a corner (gravity goes up and to the left) and pushes the boxes tightly into the corner). λ= (RG 3.1) If λi is the number of boxes in row i then λ= (λ1λ2) is a sequence of nonnegative integers and we write λn. For example, the partition in (RG 3.1) is λ= (55311) and λ15. A standard tableau of shape λ is a filling T of the boxes of λ with 1,2,...,n such that

  1. the rows of T are increasing left to right,
  2. the columns of T are increasing top to bottom.
For example, T= 1 2 5 6 10 3 4 7 12 15 8 9 14 11 16 18 13 17 is a standard tableau of shape λ=(55311). The rows and columns of the partition λ are numbered as for matrices, T(i)   is the box containing  i  in  T,  and c(b)=j-i,   if   b   is in position   (i,j)  in  λ. The number c(b) is the content of the box b. 0 1 2 3 4 -1 0 1 2 3 -2 -1 0 -3 -2 -1 -4 -5 Contents of boxes

  1. The irreducible representations Sλ of the symmetric group Sn are indexed by partitions λ with n boxes.
  2. dimSλ= # of standard tableaux of shape λ.
  3. The irreducible Sn-module Sλ is given by Sλ = -span {vT  |  T   is a standard tableau of shape   λ} with basis {vT} and with Sn action given by sivT = ( 1 c(T(i+1)) - c(T(i)) ) vT + ( 1+ 1 c(T(i+1)) - c(T(i)) ) vsiT , where
    1. si=(i,i+1),
    2. c(T(i)) is the content of the box containing i in T,
    3. siT is the same as T except that i and i+1 are switched,
    4. vviT=0 if siT is not standard.

Proof.
There are four things to show:
  1. The Sλ are Sn-modules,
  2. The Sλ are non-isomorphic,
  3. The Sλ are all irreducible,
  4. The Sλ,λn, are all irreducible Sn-modules.
  1. In order to show that Sλ is an Sn-module there are three relations to check
    1. si2=1,
    2. sisj=sjsi, if |i-j|>1,
    3. sisi+1si=si+1sisi+1.
    1. The action of si preserves the subspace S spanned by the vectors vQ indexed by the standard tableaux in the set {Q,siQ}. Depending on the relative positions of the boxes containing i and i+1 this space is either one or two dimensional. Case (A) Case (B) Case (C) In Case (A) the space S is one dimensional and si acts by the matrix M(si)=1. In Case (B), S is one dimensional and si acts by the matrix M(si)=-1. In Case (C), S is two dimensional and si acts on the subspace S by a matrix of the form M(si) = 1 a-b 1+ 1 b-a 1+ 1 a-b 1 b-a . In cases (A) and (B) it is immediate that M(si)2=1 and in case (C), since Tr(M(si))=0 and det(M(si))=-1, it follows from the Cayley-Hamilton theorem that M(si)2=Id.
    2. The action of si and sj commute if |i-j|>1 because si only moves i and i+1 in T and sj only moves j and j+1. The condition |i-j|>1 guarantees that {i,i+1} and {j,j+1} do not intersect.
    3. According to the formulas for the action, si and si+1 preserve the subspace S spanned by the vectors vQ indexed by the standard tableaux Q in the set {L,siL, si+1L, sisi+1L, si+1siL, sisi+1siL }. Depending on the relative positions of the boxes containing i, i+1, i+2 in L, this space is either 1, 2, 3, or 6 dimensional. Representative cases are when these boxes are positioned in the following ways. Case (1) Case (2) Case (3) Case (4) In Case (1) the space S is one dimensional and spanned by the vector vQ corresponding to the standard tableau
      a b c
      where a=i,b=i+1, and c=i+2. The action of Ti and Ti+1 on S is given by the matrices M(si) = (1), and M(si+1) = (1), respectively. In case (2) the space S is two dimensional and spanned by the vectors vQ corresponding to the standard tableaux
      a b c a c b
      where a=i,b=i+1, and c=i+2. The action of Ti and Ti+1 on S is given by the matrices M(si) = 1 0 0 -1 and M(si+1) = 1 2 3 2 1 2 - 1 2 . In case (3) the space S is three dimensional and spanned by the vectors vQ corresponding to the standard tableaux
      c a b b a c a b c
      where a=i,b=i+1, and c=i+2. The action of Ti and Ti+1 on S is given by the matrices M(Ti) = 1 0 0 0 1 x1-x3 1+ 1 x3-x1 0 1+ 1 x1-x3 1 x3-x1 and ϕS(Ti+1) = 1 x2-x3 1+ 1 x3-x2 0 1+ 1 x2-x3 1 x3-x2 0 0 0 1 respectively, where x1 = c(L(i)),   x2 = c(Li+1)) and x3 = c(L(i+2)). In case (4) the space S is six dimensional and spanned by the vectors vQ corresponding to the standard tableaux
      c b a c a b b c a a c b b a c a b c
      where a=i,b=i+1, and c=i+2. The action of Ti and Ti+1 on S is given by the matrices M(si) = 1 x1-x2 1+ 1 x2-x1 0 0 0 0 1+ 1 x1-x2 1 x2-x1 0 0 0 0 0 0 1 x1-x3 1+ 1 x3-x1 0 0 0 0 1+ 1 x1-x3 1 x3-x1 0 0 0 0 0 0 1 x2-x3 1+ 1 x3-x2 0 0 0 0 1+ 1 x2-x3 1 x3-x2 and M(si+1) = 1 x2-x3 0 1+ 1 x3-x2 0 0 0 0 1 x1-x3 0 0 1+ 1 x3-x1 0 1+ 1 x2-x3 0 1 x3-x2 0 0 0 0 0 0 1 x1-x2 0 1+ 1 x2-x1 0 1+ 1 x1-x3 0 0 1 x3-x1 0 0 0 0 1+ 1 x1-x2 0 1 x2-x1 where x1 = c(L(i)),   x2 = c(L(i+1)) and x3 = c(L(i+2)). In each case we compute directly the products M(si) M(si+1) M(si) and M(si+1) M(si) M(si+1) and verify that they are equal.
  2. Let T be a standard tableau with n boxes and let T- be the standard tableau T except with the box containing n removed. Then T is the unique standard tableau given by adding a box (containing n) to T- in the diagonal of boxes with content c(T(n)). Thus, by induction on the number of boxes in T, T   is uniquely determined by the sequence   (c(T(1)),c(T(2)),...,c(T(n))). For example: if (c(T(1)),c(T(2)),...,c(T(n))) = (0,1,-1,0,2,-2,-1,-3,1,3,2,-4,-3,-5,-6) then T= 1 2 5 10 3 4 9 11 6 7 8 13 12 14 15 It follows that the vector vT is, up to constant multiples, the unique vector in all the Sλ such that xkvT = c(T(k))vT, for all 1kn. This proves that the Sλ are nonisomorphic.
  3. Let us show that Sλ is irreducible. Let N be a nonzero submodule of Sλ and let n=TaTvT be a nonzero vector in N. Fix T such that aT0. It follows from the lemma and the fact that T is determined by the sequence (c(T(1)),...,c(T(n))) that the element in Sn given by pT = k=1n jc(T(k)) -njn xk-j c(T(k))-j , acts on  Sλ  by pTvS = δSTvT, for all standard tableaux S of shape λ. So (1/aT)pTn = vTN.

    Let Q be a standard tableau of shape λ and let k be minimal such that k+1 is southwest (strictly south and strictly west) of k in Q.
    k k+1
    In k exists then skW is a standard tableau and if k doesn't exist then Q is the column reading tableau, i.e. Q is given by filling the boxes of λ sequentially down columns from left to right. C= 1 6 9 12 14 2 7 10 13 15 3 8 11 4 5 Column reading tableau By repeating the procedure with skQ in place of Q repeatedly we construct a sequence Q,sk1Q, sk2sk1Q,..., skrsk2sk1Q = C, which ends at the column reading tableau C. By concatenating this sequence with a similar sequence for T, T, sl1T, ..., slmsl2sl1T = C we obtain a sequence T, sl1T, ..., C,..., sk1Q, Q = sk1skrslmsl1T of standard tableaux of shape λ which begins at T and ends at Q.

    If P is a standard tableau in this sequence and vPN and skP is the next standard tableau in the sequence, then ( sk- 1 c(P(k+1))-c(P(k)) )vP = ( 1+ 1 c(P(k+1))-c(P(k)) )vP = ( c(P(k+1))-c(P(k))+1 c(P(k+1))-c(P(k)) )vskP. Since skP is standard, c(P(k+1)) - c(P(k)) +10 and so it follows that vskPN. Thus since vTN it follows that vQN.

    So vQN for all standard tableaux of shape λ. So N contains a basis of Sλ. So N=Sλ. So N is irreducible.
  4. Let fλ be the number of standard tableaux of shape λ. Then n! = λn(fλ)2. The proof is accomplished by giving a bijection between permutations   wSn and pairs   (P,Q)   of standard tableaux, of the same shape, with  n  boxes. The matching of wSn with a pair (P,Q) is accomplished by a recursive algorithm which begins with (P0,Q0) = (,) and, at step i, "inserts" w(i) into (Pi-1,Qi-1) to obtain (Pi,Qi). The output of the algorithm is the pair of tableaux (P,Q) = (Pn,Qn).

    The insertion step for inserting w(i) into (Pi-1,Qi-1):
    1. Place w(i) into the first column of Pi-1 as follows. If w(i) is greater than all entries of column 1 of Pi-1 then place w(i) at the end of column 1. Otherwise there is a unique entry in column 1 which can be replaced by w(i) and preserve the standardness condition. (This is the smallest entry larger than w(i).)
    2. If w(i) displaces an entry from column 1 then the displaced entry is inserted into column 2 (in exactly the same way that w(i) was inserted into column 1). The displaced entries are successively inserted into the next column, and the process continues until there is no displaced entry.
    The resulting standard tableau is Pi, which contains one more box than Pi-1. The tableau Qi is obtained by adding a box filled with i to Qi-1 so that Pi and Qi are the same shape.

    To show that the algorithm produces a bijection it is necessary to confirm that the process can be reversed. To see this note that, given (Pi,Qi), the position of the box containing i in Qi tells which box of Pi must be "uninserted" to obtain Pi-1. The column by column insertion of this box will bump out an entry from column 1 of Pi and this entry is w(i) for the permutation w which corresponds to the pair (P,Q).

Define elements xk, 1kn, in the group algebra Sn of the symmetric group by x1=0, and xk = i<k (i,k), 2kn, where (i,k)Sn is the transposition which switches i and k. The action of xk on the Sn-module Sλ is given by xkvT = c(T(k))vT, for all standard tableaux   T, where c(T(k)) is the content of the box containing k in T.

Proof.
The proof is by induction on k using the relation xk = sk-1 xk-1 sk-1 + sk-1. Clearly, x1vT = 0 = c(T(1))vT for all standard tableaux T. By the induction assumption and the definition of the action of sk-1 on Sλ, xkvT = (sk-1xk-1sk-1+sk-1)vT = sk-1 (xk-1sk-1+1)vT = sk-1 ( c(T(k-1)) c(T(k))-c(T(k-1)) vT + c(T(k)) ( 1+ c(T(k-1)) c(T(k))-c(T(k-1)) )vsk-1T +vT ) = sk-1 ( c(T(k)) c(T(k))-c(T(k-1)) vT + c(T(k)) ( 1+ c(T(k-1)) c(T(k))-c(T(k-1)) )vsk-1T +vT ) = c(T(k)) sk-1 sk-1 vT = c(T(k))vT.

Notes and References

Where are these from?

References

References?

page history