ebook img

Characters and inversions in the symmetric group PDF

9 Pages·0.106 MB·English
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Characters and inversions in the symmetric group

CHARACTERS AND INVERSIONS IN THE SYMMETRIC GROUP Anne de M(cid:19)edicis, Victor Reiner, Mark Shimozono Department of Mathematics University of Minnesota Minneapolis, MN 55455 November 1993 Abstract We consider sums of the form (cid:31)(cid:21)=(cid:22)((cid:25)) qinv((cid:25)) (cid:25)X2Sn where (cid:31)(cid:21)=(cid:22) is a skew character of the symmetric group and inv((cid:25)) is the number of inversions of (cid:25). Our main result gives a lower bound on the number of factors of 1+q and 1(cid:0)q which dividethe abovesum, and isshownto be sharpwhen (cid:21)=(cid:22) is a hook partition shape. R(cid:19)esum(cid:19)e Nous consid(cid:19)eronsdes sommes de la forme (cid:31)(cid:21)=(cid:22)((cid:25)) qinv((cid:25)) (cid:25)X2Sn ou(cid:18) (cid:31)(cid:21)=(cid:22) estuncaract(cid:18)eregauchedu groupesym(cid:19)etriqueetinv((cid:25)) d(cid:19)esignelenombre d’inversions de (cid:25). Notre r(cid:19)esultat principal donne une borne inf(cid:19)erieure pour le nombredefacteurs (1+q)et (1(cid:0)q)quidivisentcessommes. Nousd(cid:19)emontronsen particulier que cette borne est exacte lorsque (cid:21)=(cid:22) est une(cid:19)equerre. TypesetbyAMS-TEX 1 2 ANNE DE ME(cid:19)DICIS, VICTOR REINER, MARK SHIMOZONO Section 1. Introduction In this paper, we consider sums over the symmetric group S of the form n (cid:31)(cid:21)=(cid:22)((cid:25))qinv((cid:25)) (cid:25)X2Sn where (cid:31)(cid:21)=(cid:22) is a skew character of the symmetric group [JP] and inv((cid:25)) is the number of inversions of (cid:25), i.e. inv((cid:25))=#f(i;j):1(cid:20)i<j (cid:20)n; (cid:25)(cid:0)1(i)>(cid:25)(cid:0)1(j)g: There are two motivations for considering such sums. Firstly, other than some special results of Gessel [Ge] and Edelman [Ed], there is very little known about the joint distribution of inversions and conjugacy class (i.e. cycle type) in S . n Thisisinstarkcontrasttothe jointdistributionsknownforinversionsanddescent statistics [GaGe], and for cycle type and descent statistics [GR]. Thus it seems reasonable to ask what can be said about the sum f((cid:25))qinv((cid:25)) (cid:25)X2Sn whenf issomeclassfunction,i.e. afunctionwhichisconstantonconjugacyclasses in S , such as an irreducible character (cid:31)(cid:21). n Remark As pointed out by the referee, there is a di(cid:11)erent interpretation one can attach to these sums, namely that (cid:31)(cid:21)=(cid:22)((cid:25))qinv((cid:25)) =hs ;F i (cid:21)=(cid:22) n (cid:25)X2Sn where h;i is the usual Hall inner product on symmetric functions, s is the skew (cid:21)=(cid:22) Schur function correspondingto (cid:21)=(cid:22), and F is a certain symmetric function with n coe(cid:14)cients in Z[q] which encodes all the information about the distribution of inversions and conjugacy class in S : n F := qinv((cid:25))p n (cid:21)((cid:25)) (cid:25)X2Sn where p ((cid:25)) is the power sum symmetric function corresponding to the cycle type (cid:21) of (cid:25) (see [Mac]). Thesecondmotivationis byanalogyto the workof [DF,Re]. Hereit wasshown that sums of the form (cid:31)(w)qdes((cid:25)) (cid:25)2W X have high divisibility by linear polynomial factors, where W is a classical Weyl group, (cid:31) is a one-dimensional character of W, and des is the descent statistic on W. Our main theorem is the following CHARACTERS AND INVERSIONS IN THE SYMMETRIC GROUP 3 Theorem 1. Let (cid:21)=(cid:22) be a skew shape with longest row of length r and longest column of length s. Then (cid:31)(cid:21)=(cid:22)((cid:25)) qinv((cid:25)) (cid:25)X2Sn is divisible by (1+q)br=2c(1(cid:0)q)bs=2c: This Theorem is proven in Section 2. The divisibilities in the previous theorem are sometimes tight, as demonstrated by Theorem 2. If (cid:21) is a hook partition (r;1s(cid:0)1) then (cid:31)(cid:21)((cid:25)) qinv((cid:25)) =(1(cid:0)q)bs=2cS(q) (cid:25)X2Sn =(1+q)br=2cR(q) where S(q);R(q)2Z[q] satisfy (n+1)!bs=2c! s is even S(1)= (s+1)! ( n!bs=2c! s is odd s! and (n+1)!br=2c! r is even R((cid:0)1)= (r+1)! ( n!br=2c! r is odd r! This is proven in Section 2. It is also conjectured there that S(q) has non- negative coe(cid:14)cients whenever r (cid:21) s. In Section 3, we look more closely at the very special case s=2 where can provethis conjecture by an interesting bijection. In this special case, the results relate to the joint distribution of (cid:12)xed points and inversions overthe symmetric group. Section 2. Proof of Theorem 1 Theorem 1 will follow by a sequence of straightforward reductions from the following lemma, which appears to be the essence of the divisibilities appearing in this context. Notation: for A(cid:18)[n], let S be the subgroup of permutations in S A n which only permute within the elements of A and (cid:12)x all elements of [n](cid:0)A. Lemma 3. Fix a subset A (cid:18) [n] having cardinality r and (cid:12)x any permutation (cid:27) 2S . Then [n](cid:0)A qinv((cid:25)(cid:27)) (cid:25)X2SA is divisible by (1+q)br=2c Proof:. Make a change of variable to p=1+q. We need to show pbr=2c divides (p(cid:0)1)inv((cid:25)(cid:27)) = p#J((cid:0)1)inv((cid:25)(cid:27))(cid:0)#J (cid:25)X2SA (cid:25)X2SA J(cid:18)Inv((cid:25)) = ((cid:0)1)inv((cid:27))+#Jp#J ((cid:0)1)inv((cid:25)) J(cid:18)X([n2]) J(cid:18)(cid:25)XI2nSvA((cid:25)) 4 ANNE DE ME(cid:19)DICIS, VICTOR REINER, MARK SHIMOZONO Here Inv((cid:25)) denotes the inversion set of (cid:25), that is Inv((cid:25))=f(i;j):1(cid:20)i<j (cid:20)n; (cid:25)(cid:0)1(i)>(cid:25)(cid:0)1(j)g So it su(cid:14)ces to show that for any J (cid:18) [n] with #J <br=2c we have 2 (cid:16) (cid:17) ((cid:0)1)inv((cid:25)) =0 (cid:25)X2SA J(cid:18)Inv((cid:25)) Since 2#J (cid:20) r(cid:0)2, there must exist a pair i;j 2 A which are not involved in any of the pairsin J, andhence multiplication on the left bythe transposition(ij) isa sign-reversing involution which shows all terms in the above sum cancel. (cid:3) Remark The preceding lemma bears some resemblance to results of Bjo(cid:127)rner and Wachs [BW] on the distribution of inversions over certain subsets of S which they call n generalized quotients. In particular,for certain of these generalizedquotients, they produce nice hook-formula factorizations for the generating function of inversions, whichnaturallyhavehighdivisibilityby(1+q). Itwouldbedesirabletounderstand this connection better. Proof of Theorem 1. We wishtoshowthat if(cid:21)=(cid:22) isaskewshapewith longestrow of length r and longest column of length s, then (cid:31)(cid:21)=(cid:22)((cid:25)) qinv((cid:25)) (cid:25)X2Sn is divisible by (1+q)br=2c(1(cid:0)q)bs=2c: Our strategy will be to reduce the theorem to a very special case and then apply the Lemma. First note that if ((cid:21)=(cid:22))t denotes the skew shape which is the transpose of (cid:21)=(cid:22), then (cid:31)((cid:21)=(cid:22))t((cid:25)) qinv((cid:25)) = (cid:31)(cid:21)=(cid:22)((cid:25)) sgn((cid:25)) qinv((cid:25)) (cid:25)X2Sn (cid:25)X2Sn = (cid:31)(cid:21)=(cid:22)((cid:25)) ((cid:0)q)inv((cid:25)) (cid:25)X2Sn using the fact that (cid:31)((cid:21)=(cid:22)t)((cid:25))=(cid:31)(cid:21)=(cid:22)((cid:25)) sgn((cid:25)) Therefore it su(cid:14)ces to show only divisibility by (1+q)br=2c. Using the Littlewood-Richardson rule [Sa], one knows that (cid:31)(cid:21)=(cid:22)((cid:25))= (cid:31)(cid:23)((cid:25)) (cid:23) X CHARACTERS AND INVERSIONS IN THE SYMMETRIC GROUP 5 where the sum rangesovera multiset of partitions (cid:23) which all havelongest part of size at least r. Then using the Jacobi-Trudi identity [Sa], one can write (cid:31)(cid:23)((cid:25))= (cid:6)(cid:31)H((cid:26))((cid:25)) (cid:26) X Here (cid:26) runs over a set of partitions of n, and H((cid:26)) is the skew diagram having disjoint horizontal rows of size (cid:26) for each i. i Furthermore,expanding the Jacobi-Trudideterminant along its top rowtells us that each (cid:26) appearingin the abovesum will haveits largestpart (cid:26) of size at least 1 r. Since we are only trying to show divisibility by (1+q)br=2c, it therefore su(cid:14)ces to prove the theorem in the special case where (cid:21)=(cid:22) = H((cid:26)) for some partition (cid:26) with largest part of size r. Thecharacter(cid:31)H((cid:26)) is easytowrite downexplicitly, asit isthe characterofthe permutation representation of S acting on the left cosets of the Young subgroup n S which permutes the (cid:12)rst (cid:26) numbers among themselves, permutes the next (cid:26) (cid:26) 1 2 among themselves, etc. So (cid:31)H((cid:26))((cid:25)) is the number of such left cosets (cid:12)xed by (cid:25). Therefore, (cid:31)H((cid:26))((cid:25)) qinv((cid:25)) = #fcosets (cid:28)S :(cid:25)(cid:28)S =(cid:28)S g qinv((cid:25)) (cid:26) (cid:26) (cid:26) (cid:25)X2Sn (cid:25)X2Sn = qinv((cid:25)) coseXts(cid:28)S(cid:26) (cid:25)(cid:28)(cid:25)SX2(cid:26)=Sn(cid:28)S(cid:26) = qinv((cid:25)1(cid:1)(cid:1)(cid:1)(cid:25)k) (A1X;:::;Ak) ((cid:25)1X;:::;(cid:25)k) ]Ai=[n];#Ai=(cid:26)i (cid:25)i2SAi = qinv((cid:25)1(cid:27)) ]A(iA=1[Xn;:]:;:#;AAki=)(cid:26)i ((cid:25)(cid:25)2iX;2::S:;A(cid:25)ik)(cid:25)1X2SA1 where (cid:27) =(cid:25) (cid:1)(cid:1)(cid:1)(cid:25) in the last summation. By the Lemma, each of these sums 2 k qinv((cid:25)1(cid:27)) (cid:25)1X2SA1 are divisible by (1+q)br=2c, and hence the theorem is proven. (cid:3) Remark At least some of the divisibility asserted by Theorem 1 is immediate, namely that fors(cid:21)2the sumisdivisible by(1(cid:0)q)1,andbythe transposesymmetry,forr (cid:21)2 thesumisdivisibleby(1+q)1. Wesimplynotethatbythe Littlewood-Richardson rule,fors(cid:21)2,(cid:31)(cid:21)=(cid:22) doesnot containthe trivialcharacterinits irreducibledecom- position, and hence by the orthogonalityrelation for characters, we have (cid:31)(cid:21)=(cid:22)((cid:25)) qinv((cid:25)) = (cid:31)(cid:21)=(cid:22)((cid:25))(cid:1)1((cid:25))=0 " # (cid:25)X2Sn q=1 (cid:25)X2Sn This implies the sum is divisible by (1(cid:0)q)1. Section 3. Proof of Theorem 2 6 ANNE DE ME(cid:19)DICIS, VICTOR REINER, MARK SHIMOZONO We only give a sketch of the proof. To begin, we note that our earlier comment about the e(cid:11)ect of transposing the shape(cid:21)=(cid:22)onthesumimpliesthatweonlyneedtoprovethatS(1)hastheasserted value. We will proceed by descending induction on r, beginning with the base case r =n. Note that in this case (cid:31)(r;1s(cid:0)1) is the trivial character and qinv((cid:25)) =[n]! =[n] [n(cid:0)1] (cid:1)(cid:1)(cid:1)[2] [1] q q q q q (cid:25)X2Sn where [n] = 1+q +q2 +:::+qn(cid:0)1. The theorem then immediately follows in q this case from the aboveexpression. Forthe inductive step, ourstrategywill be to replace(cid:31)(r;1s(cid:0)1) with a characterwecan compute more explicitly. Let (1s(cid:0)1)(cid:8)(r) denote the skew shape which has a column of size s(cid:0)1 disjoint from a row of size r. By an easy case of the Littlewood-Richardson rule, (cid:31)(1s(cid:0)1)(cid:8)(r) =(cid:31)(r;1s(cid:0)1)+(cid:31)(r+1;1s(cid:0)2) and hence (cid:31)(r;1s(cid:0)1) =(cid:31)(1s(cid:0)1)(cid:8)(r)(cid:0)(cid:31)(r+1;1s(cid:0)2) Using the last equation, one can check that the inductive step is equivalent to proving the following: (cid:31)(1s(cid:0)1)(cid:8)(r)((cid:25)) qinv((cid:25)) (n+2)n!br=2c! r is even (cid:25)2Sn = (r+1)! "P (1+q)br=2c #q=(cid:0)1 ( n!brr=!2c! r is odd Toprovethelastexpression,wemakethe changeofvariablep=1+q asbefore, andwewillusethefactthatthecharacter(cid:31)(1s(cid:0)1)(cid:8)(r)istheinductionfromS (cid:2)S s(cid:0)1 r to S of the charactersgn(cid:10)1. We then have n (cid:31)(1s(cid:0)1)(cid:8)(r)((cid:25)) qinv((cid:25)) (cid:25)2Sn "P (1+q)br=2c # q=(cid:0)1 (cid:31)(1s(cid:0)1)(cid:8)(r)((cid:25)) (p(cid:0)1)inv((cid:25)) = (cid:25)2Sn "P pbr=2c # p=0 =2p(cid:0)br=2c sgn((cid:25)j )(p(cid:0)1)inv((cid:25))3 A 66 (cid:25)X2Sn A2X(s[(cid:0)n]1) 77 6 (cid:25)(A)=A 7p=0 4 5 =2p(cid:0)br=2c sgn((cid:25)j ) p#J((cid:0)1)inv((cid:25))(cid:0)#J3 A 66 (cid:25)X2Sn A2X(s[(cid:0)n]1) J(cid:18)X([n2]) 77 6 (cid:25)(A)=A J(cid:18)Inv((cid:25)) 7p=0 4 5 =((cid:0)1)br=2c sgn((cid:25)j ) [n](cid:0)A A2(s[(cid:0)n]1X);J(cid:18)([n2]) (cid:25)(A)=A(cid:25)X;2JS(cid:18)nInv((cid:25)) #J=br=2c CHARACTERS AND INVERSIONS IN THE SYMMETRIC GROUP 7 The proof then proceeds by (cid:12)nding a sequence of relatively simple sign-reversing involutions which sieve the above sum down to a small set of terms, each having sign ((cid:0)1)br=2c. and having the desired cardinality to prove the assertion. (cid:3) The data suggests that even more is true in the hook case: Conjecture 4. Let (cid:21) be a hook partition (r;1s(cid:0)1), and S(q) as before, i.e (cid:31)(cid:21)((cid:25)) qinv((cid:25)) =(1(cid:0)q)bs=2cS(q) (cid:25)X2Sn If r (cid:21)s, then S(q) has non-negative coe(cid:14)cients. Section 4. The de(cid:12)ning representation When s = 2, the character (cid:31)(r;1s(cid:0)1) is essentially the de(cid:12)ning character of S , n i.e. the permutation representation of S acting on [n]. This character has the n particularly simple expression (cid:31)(n(cid:0)1;1)((cid:25))=#Fix((cid:25))(cid:0)1 where Fix((cid:25)) is the set of (cid:12)xed points or 1-cycles of (cid:25) i.e. Fix((cid:25))=fk:1(cid:20)k (cid:20)n; (cid:25) =kg k Therefore our results in this case have interpretations in terms of the joint distri- bution of (cid:12)xed points and inversions over the symmetric group, some of which are interesting. For example, it is well-known and easy to see that \the average per- mutationinS hasone(cid:12)xedpoint"and\theaveragepermutationinS has 1 n n n 2 2 inversions". The following corollary asserts that the value of #Fix((cid:25))(cid:1)inv((cid:25)) for (cid:0) (cid:1) the \averagepermutation (cid:25) in S " is n (3n+1)(n(cid:0)2) 12 Corollary 5. (3n+1)(n(cid:0)2) #Fix((cid:25))(cid:1)inv((cid:25))= n! 12 (cid:25)X2Sn Proof. Let f(q) denote the sum (cid:31)(n(cid:0)1;1)((cid:25))qinv((cid:25)) = (#Fix((cid:25))(cid:0)1)qinv((cid:25)) (cid:25)X2Sn (cid:25)X2Sn Then we have lim f(q) S(1)= q !1 1(cid:0)q (cid:18) (cid:19) d = f(q) dq (cid:20) (cid:21)q=1 = (#Fix((cid:25))(cid:0)1)(cid:1)inv((cid:25)) (cid:25)X2Sn By Theorem 2 in the case s=2, we have S(1)=(n+1)!=6. Combining this with the easy fact that n n! inv((cid:25))= 2 2 (cid:25)X2Sn (cid:16) (cid:17) gives the result. (cid:3) The preceding result can be considerably strengthened: 8 ANNE DE ME(cid:19)DICIS, VICTOR REINER, MARK SHIMOZONO Proposition 6. n k(cid:0)1 n(cid:0)k #Fix((cid:25)) qinv((cid:25)) = [k(cid:0)1]! [n(cid:0)k]! qj2+2j q q j j (cid:25)X2Sn Xk=1 Xj(cid:21)0 (cid:20) (cid:21)q(cid:20) (cid:21)q where n [n]! q = k q [k]!q[n(cid:0)k]!q . h i Proof. n #Fix((cid:25)) qinv((cid:25)) = qinv((cid:25)) (cid:25)X2Sn Xk=1 (cid:25)(cid:25)X2k=Snk n = qinv((cid:27)) qinv((cid:28)) Xk=1 (cid:27)2S[1;k(cid:0)1](cid:2)XSfkg(cid:2)S[k+1;n] (cid:28)1<(cid:1)(cid:1)(cid:1)(cid:28)(cid:28)k2(cid:0)SX1n;;(cid:28)(cid:28)kk+=1k<(cid:1)(cid:1)(cid:1)(cid:28)n where the last equality comes from the well-known fact that any permutation (cid:25) may be factored uniquely as (cid:25) =(cid:27)(cid:28) with (cid:27) 2S (cid:2)S (cid:2)S and [1;k(cid:0)1] fkg [k+1;n] (cid:28) <(cid:1)(cid:1)(cid:1)<(cid:28) and (cid:28) <(cid:1)(cid:1)(cid:1)(cid:28) 1 k(cid:0)1 k+1 n Note that we are restricting this factorization to the set of (cid:25) which (cid:12)x k, but this creates no di(cid:14)culties. Since qinv((cid:25)) =[n]! , we have (cid:25)2Sn q P n #Fix((cid:25)) qinv((cid:25)) = [k(cid:0)1]! [n(cid:0)k]! qinv((cid:25)) q q (cid:25)X2Sn kX=1 (cid:28)1<(cid:1)(cid:1)(cid:1)(cid:28)(cid:28)k2(cid:0)SX1n;;(cid:28)(cid:28)kk+=1k<(cid:1)(cid:1)(cid:1)(cid:28)n and it only remains to compute the last sum on the right-hand side. Given such a (cid:28), de(cid:12)ne two subsets J ;J (cid:18)[n] by 1 2 J =fi:1(cid:20)i(cid:20)k(cid:0)1; (cid:28)(cid:0)1(i)(cid:21)k+1g 1 J =fi:k+1(cid:20)i(cid:20)n; (cid:28)(cid:0)1(i)(cid:20)k(cid:0)1g 2 NotethatJ ;J musthavethesamecardinalitywhichwedenotebyj,andchoosing 1 2 thesetsJ ;J completelydetermines(cid:28). Thereareexactlyj2 inversionsbetweenJ 1 2 1 andJ ,andexactly2j inversionsbetweenkandJ ;J . Thisaccountsfortheqj2+2j 2 1 2 term in the sum. The remaining inversions of (cid:28) come from among the numbers in [1;k (cid:0)1] and among the numbers in [k +1;n], and correspond to MacMahon’s \inv" statistic on subsets [Ma] applied to the sets J ;J respectively. Since the 1 2 distributionof\inv"overk-subsetsofann-setisgivenby n ,thetheoremfollows. k q (cid:3) (cid:2) (cid:3) Remark The last sum on the right-hand side in the preceding theorem is similar to the q-Vandermonde convolution k(cid:0)1 n(cid:0)k n(cid:0)1 qj2 = j j k j(cid:21)0 (cid:20) (cid:21)q(cid:20) (cid:21)q (cid:20) (cid:21)q X CHARACTERS AND INVERSIONS IN THE SYMMETRIC GROUP 9 but does not sum so nicely. However, when we set q = 1 or q = (cid:0)1, the two are the same. Finally, we wish to prove the s = 2 case of Conjecture 4. Unravelling the statement in this case, it conjectures that for all k (cid:20)n, one has #Fix((cid:25))(cid:21)#f(cid:25) 2S :inv((cid:25))(cid:20)kg n in(cid:25)vX2((cid:25)S)n(cid:20)k Recalling that the average permutation in S has one (cid:12)xed point, this may be n paraphrasedas sayingthat the set of all permutations with length at most k have, on average, \more than their fair share" of (cid:12)xed points. This seems plausible, sinceonewouldthinkthathavingfewinversionsandhaving(cid:12)xedpointsshouldbe positively correlated. It is not hard to see that the assertion follows from Theorem 7. There exists a bijection (cid:30) between the sets f((cid:25);k):(cid:25) 2S ; k 2[n]; (cid:25) =kg$S n k n with the property that if (cid:30)((cid:25);k)=(cid:27) then inv((cid:27))(cid:21)inv((cid:25)). Proof. First de(cid:12)ne the folding bijection f :[n]![n] by 2 minfk(cid:0)1;n(cid:0)kg+1 if k (cid:20)dne f(k)= 2 2 minfk(cid:0)1;n(cid:0)kg+2 if k >dne (cid:26) 2 Given a pair ((cid:25);k) as in the left-hand side of the theorem, de(cid:12)ne the permutation (cid:27) = (cid:27) (cid:27) (cid:1)(cid:1)(cid:1)(cid:27) as follows: (cid:27) = f(k), and (cid:27) ;::: ;(cid:27) are the numbers [n](cid:0)f(k) 1 2 n 1 2 n listed in the same relative order of magnitude as (cid:25) ;::: ;(cid:25) ;(cid:25) ;:::(cid:25) . It is 1 k(cid:0)1 k+1 n easy to check that this is a bijection with the desired property. (cid:3) Cansuchabijection(cid:30)befoundsothat(cid:27) (cid:21)(cid:25) inweak Bruhatorder? Thiswould imply that the permutations in any lower order ideal of weak Bruhat order have, on average,\more than their fair share" of (cid:12)xed points. Section 5. Acknowledgements The authors would like to thank Paul Edelman and Dennis Stanton for helpful discussions and references. References [BWA].Bjo(cid:127)rnerandM.Wachs,Generalized quotientsinCoxeter groups,Trans.AMS308,1-37. [Ed]P.H.Edelman,On inversionsand cyclesinpermutations,Eur.J.Comb.8,269-279. [DF]J.D(cid:19)esarm(cid:19)enienandD.Foata,Thesigned Eulerian numbers,preprint. [GGA] .GarsiaandI.Gessel,Permutation statisticsand partitions,Adv.Math.31,288-305. [Ge]I.Gessel,A q-analogue ofthe exponential formula,Disc.Math.40,69-80. [GRI]. Gessel and C. Reutenauer, Counting permutations by descent set and cycle structure, preprint. [Ma]P.A.MacMahon,Combinatory Analysis, Vol.1,CambridgeUniv.Press,Cambridge. [Re]V.Reiner,Descents and one-dimensional characters for classical Weyl groups, to appear in Disc.Math.. [Sa]B. Sagan, The symmetric group: representations, combinatorial algorithms, and symmetric functions,Wadsworth&Brooks/Cole,Paci(cid:12)cGrove,1991.

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.