Quiver Hecke Algebras

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

Last updates: 27 July 2012

Quiver Hecke Algebras Rα

Let C= (αi, αj) be a symmetrizable Cartan matrix so that αi, αi =2, αi, αj 0, and αi, αj =0 αj, αi =0. Let αi = 2 αi, αi αi, so that ,: 𝔥*×𝔥* , is a symmetric bilinear form on 𝔥*=span {α1,, αn}.

The parameters tij and tijrs for the quiver Hecke algebras Rα are specified by a choice of polynomials Qij (u,v)such that Qji(u,v) = Qij(v,u), Qii(u,v) =0, and Qij(u,v) = tij, if ij and αi, αj =0, and Qij(u,v) = tij u-αi, αj + tij v-αj, αi + 0r- αi, αj-1 0s- αj, αi-1 tijrs urvs, if ij and αi, αj0.

Let be the free algebra generated by f1, ,fn. The set of words in the letters f1,,fn ,

Γ*={ fi1 fid | 1 i1,,id n }, is a basis of
and the algebra is graded by
Q+ = i=1n 0αi with deg(fi) =αi.

Fix αQ+ and let Γα= {u=u1ud | uΓ*, deg(u) =α}, so that d=l(u) is the length of the words in Γα. The symmetric group Sd, generated by the simple transpositions s1,, sd-1, acts on the set Γα by permuting the positions of the letters in the words.

The quiver Hecke algebra the -graded algebra Rα given by generators eu, x1eu,, xdeu, τ1eu,, τd-1eu, for uΓα, with degrees deg(eu)=0, deg(xieu) =ui, ui, deg(τieu) = -ui, ui+1 (where ui denotes the ith letter of the word u and we write u,v for deg(u), deg(v)),   and relations euev = δuveu , xieu =euxi , τieu =esiu τi , xixj =xjxi, τi2eu =Qui ui+1 (xi, xi+1) eu, τiτj =τjτi if j i,i±1, ( τi+1 τiτi+1 -τiτi+1 τi) eu = { 1 xi+2-xi ( Q ui+2, ui+1 (xi+2, xi+1) - Q ui, ui+1 (xi, xi+1 ), if ui=ui+2, 0, otherwise, and τixjeu = { xsi(j) τieu -eu, if ui=ui+1 and j=i, xsi(j) τieu+eu , if ui=ui+1 and j=i+1, xsi(j) τieu, otherwise, where xi=xi1 =u Γα xieu, and τi=τi1 =u Γα τieu, since 1=u Γα eu in Rα.

[KL, thm. 2.5], [Ro, thm. 3.7] As in (1.2) fix αQ+, let Γα be the set of words of degree α, and let d be the length of words in Γα. The algebra Rα has a basis { τσ x1n1 xdnd eu | uΓα, σSd, n1,, nd0 } where for each σSd we fix a reduced word σ=si1 sil and set τσ= τi1 τil.

The Quiver Hecke algebra for type A

The KLR Quiver Hecke algebra  Rα for type  A is given by generators y1,,yd, eu, forud, ψ1,,ψd-1 with relations yrys = ysyr, euev = δuveu, 1=ud eu euyr = yreu, euψr = ψreu, ψrys = ysψr, if sr,r+1 ψr yr eu = yr+1 ψr - 1 eu , if ur ur+1 = ur ur , yr+1 ψr eu , otherwise , ψr yr+1 eu = yr ψr + 1 eu , if ur ur+1 = ur ur , yr+1 ψr , otherwise , ψr ψs = ψs ψr if sr, r±1, ψr2 eu = ± yr+1 - yr eu , if ur ur+1 = ur ur±1 , 0 , if ur ur+1 = ur ur , eu , otherwise , ψr ψr+1 ψr eu = ψr+1 ψr ψr+1 ±1 eu , if ur ur+1 ur+2 = ur ur±1 ur , ψr+1 ψr ψr+1 eu , otherwise ,

In this case the Cartan matrix, Dynkin diagram and quiver Hecke parameters are

C=0-12-1-12-1-12-10, -3 -2 -1 0 1 2 3 A Q=1u-v0v-uu-v0v-uu-v0v-u1

These formulas have been checked against [KR1, (2.9) and (2.11)], [KR2, (3.1) and (3.2)] and [MH, §3.1].

The Quiver Hecke algebra for C=2-m-m2

Special cases of this are
type  A 2  with  C = 2 -m -m 2 and type  A 1 1  with  C = 2 -2 -2 2 .

The Cartan matrix, Dynkin diagram and quiver Hecke parameters are

C=2-m-m2 m Q=0v-umu-vm0

The KLR Quiver Hecke algebra Rα for C=2-m-m2 is given by generators y1,,yd, eu,foru01d, ψ1,,ψd-1 with relations yrys=ysyr, euev=δuveu, 1=u01deu euyr=yreu, euψr=ψresru, ψrys=ysψr, if sr,r+1, ψr2 eu = yr+1 - yr eu , if ur u r+1 = 01 , yr - yr+1 eu , if ur u r+1 = 10 , 0 otherwise ψrψr+1ψreu= ψr+1 ψr ψr+1 - j=0m-1 yr+1 - yr+2 m-1-j y r+1 - yr j eu , if ur ur+1 ur+2 = 010 , ψr+1 ψr ψr+1 + j=0m-1 yr+2 - yr+1 m-1-j yr - y r+1 j eu , if ur ur+1 ur+2 = 101 , ψr+1 ψr ψr+1 eu , otherwise , since Q 01 yr+2 yr+1 - Q 01 yr yr+1 yr+2 - yr = yr+1 - yr+2 m - yr+1 - yr m - yr+1 - yr+2 - yr+1 - yr = - j=0 m-1 yr+1 - yr+2 m-1-j yr+1 - yr j and Q 10 yr+2 yr+1 - Q 10 yr yr+1 yr+2 - yr = yr+2 - yr+1 m - yr - yr+1 m yr+2 - yr+1 - yr - yr+1 = j=0 m-1 yr+2 - yr+1 m-1-j yr - yr+1 j

The KLR quiver Hecke algebra for C=2-m-12

This case comes from the quiver

I˜= 1 2 3 4 5 m m-1 a a a a a a with the automorphsim acyclically permuting thevertices 1,2,m.

In this case, I= 0 1 with

α0 α0 = 2·1=, α0 α1 =-m, α1 α0 = -m, α1 α1 = 2·m = 2m

d01 =  number of orbits of  a  on edges from  0  to  1=1 d10 =  number of orbits of  a  on edges from  1  to  0=0 and Q01 = -1 d01 u 2m2 - v 2m2m -2-m 2m = - um-v 1 = vum so that C= 2 -m -1 2 , A= 2 -m -m 2m , Q= 0 v-um u-vm 0

These computations are checked against [KR2, (3.1)] and [R, §3.2.4] and [KR2, (3.2)].

Special cases are

Type A2 with C= 2 -1 -1 2 , A= 2 -1 -1 2 , Q= 0 v-u u-v 0 Type B2 with C= 2 -2 -1 2 , A= 2 -2 -2 4 , Q= 0 v-u2 u-v2 0 Type G2 with C= 2 -3 -1 2 , A= 2 -3 -3 6 , Q= 0 v-u3 u-v3 0 Type C2 with C= 2 -1 -2 2 , A= 4 -2 -2 2 , Q= 0 v2-u u2-v 0

Rouquier [R, §3.2.4] follows the step of quivers with compatible automorphism as presented in [Lu, §12.1.1] so that I˜ E˜  is a directed graph with an automorphism a. Set  IE = I˜ E˜ /a and, for vertices i and j of I, αi αi = 2·(number of elements of i), αi αj = -(number of edges of E˜ between i and j), dij = -number of orbits of a on edges from i to j, and Qij = -1 dij u lcm αi αi αj αj αi αi - v lcm αi αi αj αj αj αj -2 αi αi lcm αi αi αj αj

Notes and References

These notes are partly based on joint work with A. Kleshchev.

In [Ro, Theorem 3.7], it is explained that Rγ has basis as given in Theorem 1.1 if and only if Rγ satisfies PBW if and only if Qji (u,v) =Qij (v,u).

References

[KL] M. Khovanov and A. Lauda, A diagrammatic approach to categorification of quantum groups I, Representation Theory 13 (2009), 309–347. MR2525917

[Ro] R. Rouquier, 2 Kac-Moody algebras, arXiv:08125023

[MH] A. Mathas and J. Hu, arXiv:0907.2985

page history