J. Scott Carter Daniel E.

Flath
and i~lasahico Saito

Mathematical Notes 43

1 Introduction
2 Representations of U(sZ(2))

Basicdefinitions

......................

7 7 12 15 21 27 43 63

Finite dimensional irreducible representations

...... Diagrammatics of U(sl(2)) invariant maps . . . . . . . . The Temperley-Lieb algebra . . . . . . . . . . . . . . . . Tensor products of irreducible representations . . . . . . The 6j-symbols . . . . . . . . . . . . . . . . . . . . . . . Computations . . . . . . . . . . . . . . . . . . . . . . . .
h rccu~sio~l formula for the 61-symbols . . . . . . . . . .
Iteillh~hs . . . . .

3

Quantum sZ(2)
Some finite dimensional representations Representations of the braid groups

67

.........

67

. . . . . . . . . . . 70 A finite dimensional quotient of C [ B ( n ).] . . . . . . . . 74 ............... The quantum Clebsch-Gordan theory . . . . . . . . . . Quantum network evaluation . . . . . . . . . . . . . . . The quantum 6j-symbols - generic case . . . . . . . . . Diagrammatics of weight vectors (quantum case) . . . . Twisting rules . . . . . . . . . . . . . . . . . . . . . . . .
The Jones-Wentzl projectors

80
93 99 106 110
111

Symmetries . . . . . . . . . . . . . . . . . . . . . . . . . 123 Further identities among the quantum Gj-symbols . . . 125

4

The Quantum Trace and Color Representations

127
127 130 133 139 This book discusses the representation theory of classical and quantum U ( d ( 2 ) ) with an eye towards topological applications
of the latter. We use the Temperley-Lieb algebra and the Van-

II

.............. ..... A bilinear form on tangle diagrams . . . . . . . . . . . . Color representations . . . . . . . , . . . . . . . . . . . The quantum Gj-symbol- root of unity case . . . . . .
The quantum trace . .

5 The Turaev-Viro Invariant

I

. . . . . . .. 151 Epilogue . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
The definition of the Turaev-Viro invariant

151

i

References

160

I

tum spin-networks to organize the computations. We define the 6j-symbols in the classical, quantum, and quantum-root-of-unity cases, and use these computations to define the Turaev-Viro invariants of closed 3-dimensional manifolds. Our approach is elementary and fairly self-contained. We develop the spin-networks from an algebraic point of view.

1

Introduction

These notes grew out of a series of seminars held at the University of South Alabama during 1993 that were enhanced by regular email among the three of us. We became interested in quantum diagrammatic representation theory following visits from Ruth Lawrence and Lou Kauffman to Mobile.

We develop the Clebsch-Gordan theory and the recoupling theory for representations of classical and quantum U(sl(2)).via the spin networks of Penrose 1271 and Kauffman [16]. In these theories, the finite dinlensio1la.1irreducible representa.tions are reali ~ e d spaces oi i~oluogcr~eoil:, i~oi~ii,il\ ill l,oi> 111
1\\(,
\ .I:

I . I I ) I ~ ~ ~TII

the quantum case the variables commute up to a factor of q; z.e. yx = qxy. The tensor product of two representations is decomposed as a direct sum of irreducibles, and the coefficients of the various weight vectors are computed explicitly. In the quantum case, when the parameter is a root of unity, we only decompose the representations modulo those that have trace 0. We use the spin networks to develop the theory in the classical case for two reasons. First, they simplify and unify many of the tricky combinatorial facts. The simplification of the proofs is nowhere more apparent than in Theorem 2.7.14 where a plethora of identities is proven via diagram manipulations. Second, the spin networks are currently useful and quite popular in the quantum case (see for example [23], [18], [28]). One of our goals here is to explain the representation theory of quantum sl(2) in the spin network framework. We know of no better explanation than

=I

1
I
I

-

--..--I-I-.. ..I.u

xu,II, I

UIYI

VJ-31

IVI,,"ba

IN L

~ U U U ~ I I V I V

I1

to run through the classical case (which should be more familiar), and then t o imitate the classical theory in the quantum case. Here we give an overview. The set of (2 by 2) matrices of determinant 1 over the complex numbers forms a group called SL(2). The finite dimensional irreducible representations of SL(2) are well understood. In particular, it is known how to decompose

on triangulations of a 3-manifold while the orthogonality condition can be interpreted as a Matveev [25] move on the dual 2-skeleton of a triangulation. The Turaev-Viro invariants were based on work of Kirillov and Reshetikhin on the representation of quantum groups [19]. This work together with Reshetikhin-Turaev [29] formed a mathematicdy rigorous framework for the invariants of Witten [34]. Meanwhile Kauffman and Lins [18] gave a simple combinatoric approach to the invariants based on the Kauffman bracket and the spin networks of Penrose [27]. Piunikhin [28] showed that the Kauffman-Lins approach and the Turaev-Viro approach coincide. Some of Kauffman's contributions to the subject can also be found in the papers [14], [15], and [17]. A more traditional algebraic approach to quantum groups can be found in [30]; in particular, they discuss from the outset the Hopf-algebra structures.

I

I
1

II

the tensor product of two such representations into a direct sum of irreducibles. In this decomposition one can compute explicitly the image of weight vectors and such computations form the heart of the so-called Clebsch-Gordan theory. The finite dimensional representations of S L ( 2 ) are the same as those of U(sl(2)) which is an algebra generated by symbols E, F and H subject to certain relations. Furthermore, the tensor product of three representations can be decomposed in two natural ways. The comparison of these
two decomytosit ions is sornctimcs c,rllctl r r r n ~ ~ , , l ? r i t/)ro,~ll.c l l l f l q

II

T icl;ol.isll'q [23] d(\finition of tlic R t ~ s l ~ c t i k l i i ~ l - ~iilvalainlltb ~~. ov
1s of a colllbi~~ator~al nature. The I<auiinian-L~ns [IS] d e l i n ~ t ~ 01 ~ l o the Turaev-Viro invariants is defined similarly. Neither of these combinatorial approaches relied on representation theory. However, the remarkable feature of quantum topology is that there are close connections between algebra and topology that were heretofore unimagined. The purpose of this paper is to explore these relations by examining the algebraic meaning of the diagrams and by using diagrams to prove algebraic results. Here is our outline. Section 2 reviews the classical theory of of U(sZ(2)). There is nothing new here, but we how the Clebsch-Gordan coefficients and the 6j-symbols do are computed in terms of the bracket expansion (at A = 1). In Section 3 we mimic these constructions to obtain the quantum Clebsch-Gordan and 6j-symbols. In Section 4 we will define the

..
-p

tile rccou1)llllg

t

uell~c~erit:, c. I , I ~ o \ L I ~ ,ir

LJ~C

G ~ ' - ~ ~ ~ l ~ ] ~ ~ l ~ , rlllese

symbols satisfy two fundamental identities (orthogonality and the Elliott-Biedenharn identity) that can be interpreted in terms of the decomposition of the union of two tetrahedra. In the ~ l l i ~ t t Biedenharn identity the tetrahedra are glued along a single face and recomposed as the union of three tetrahedra glued along an edge- For orthogonality the tetrahedra are glued along two faces, and the recomposition is not simplicial. The symmetry of the 6j-symbols and their relationship to M ~ a h e d r awas for the most part a mystery, until & - a e v and Viro 1321 constructed 3-manifold invariants based on the analogous theory for quantum sZ(2). The identities satisfied by the Gj-symbols are also satisfied by their quantum analogues. The Elliott-Biedenharn identity corresponds to an Alexander [I] move

I*
$1

l It

I
11
I

I/

I

"

r

V U n L J L I L U f i U

fl1lU

WUXIY 1

U M tJJ->YMB(JLS

I

/'i

quantum trace and discuss the recoupling theory in the root of unity case. Section 5 reviews the definitions of the Turaev-Viro invariants and proves that the definition is independent of the triangulation by means of the Pachner Theorem [26].

2

Representations of U ( s l ( 2 )
Let C denote the complex numbers. The

I!
I/

2.1 Definition.

group SL(2) is defined to be SL(~)=

ii

Acknowledgments. We all are grateful to L. Kauffman and R. Lawrence for the interesting conversations that we have had.
Their visits to Mobile were supported by the University of south ~ ~ ~ b a mArts and Sciences Support and Development Fund. a's Additional financial support was obtained from Alabama E P S C ~ R for funding of travel for the first named author and support of a Conference in Knot Theory, Low Dimensional Topology, and Quantum Groups in Mobile in 1994. C. Pillen, B. Kuripta, K. Murasugi, and R. Peele provided us with valuable information. Jim Stasheff read a preliminary version of the text and provided us with many helpful comments. Cameron Gordon's past finanical support of Masahico Saito was greatly appretiated. Finall!. ive all grdtcfull) dcl\nowledgc t i l e s u p p o ~ t aild patle~~cc~ L have shown to us over the years.
~ A L otll

{(; ;)
\

:a,b,c,dEC, a d - b c = l

I

\

:(/I

where the law of composition is matrix multiplication. The associated Lie algebra sZ(2) consists of the set of matrices of trace
0:

,,(,)={(;

i)

:a,b,c,di(:,

a+d=n

' 11

This is spanned by E =

wlbes

CL 11 -8 1

I

13-1.,oti,at[1:.1]=~~i.

y2 ) (
-

( ;),.=(
\r!.~]=!

I

=

a

. The Lie b m d e i is c~mputed [A, Bl = AB via

-

>-(liu 1-1----l7

exponential function, exp : sl(2) -+ SL(2), which is defined by the
j

power series:

/

$1

~ X P Q= j=o

" Q3

for Q E sZ(2). The function exp maps a trace 0 matrix to a matrix
I

with determinant 1.
2.2- Finite dimensional irreducible representations.

The

group SL(2) acts on the vector space of linear combinations of. variables x and y by

AND

QUANTUM 6j-SYMBOLS representation,
= C . The index j is sometimes called the spin

(
I

VO

;)y=cxdy

of the representation V3. The associated Lie algebra d(2) acts on V 3 8s ~ ~ U O W S :

where the action is extended linearly. This is called the fundamental representation of SL(2).

E~~~~ =

dl
dt
t=o

exp (tE)xTys = srTt1 ys-',

i
i

More generally, define an action of SL(2) on the space of polyand 1 and = (ax

F~~ = rxT-I ySC1, yS
r-s HSTyS = xTyS. 2

1 11 ' i
11

A weight vector is an eigenvector under the action of H in any
One way to verify that this is a group action is to consider the embedding representation; its eigenvalue is called its weight. For example xTyS E V ( T + S ) I ~ a weight vector of weight is the set of weights in
~j

I

t

y.Observe that

is {j, 1,j-2,. . .,- j ) , and by definition j-

the corresponding weight vectors form a basis for VJ.

.

.,

Y and where the sum is over all

2.2.1 well ~ l z o w nXheorenl. ( S e e [g] o r [ 3 3 ] , h - ~ l m ic

The representations of SL(2) On V 3 are irreducible.

geneous ~ o l ~ n o m i aof degree ( r ls

+ s ) in x and y into the tensor

proof.1f W is an SL(2)-subrepresentation of V3, then W is also
invariant under the action of the algebra sl(2) that is given above. Therefore, it is enough to show that the representation of sl(2) On

product of ( r + s ) copies of the fundamental represelltation space.

The tensor product V 8 W of representations V and ]/V illherits
an action via g(v @ W ) = gv 8g w where v E V and w E W . ~h~~ if

vj is irreducible.
The matrix E acts by sending a weight vector to one of higher

I denotes the fundamental representation space, then v@(T+s)is '
also arepresentation space. Furthermore, the image of the space of ~olynomials consists of the subspace of tensors that fixed under the! action of the permutation group on the tensor factors of VB('+"), and this space is stable under the action of SL(2). It is customary to let nomials of degree 2 j = r
~j

F sends such to one of lower weight. Since the image weight of anv non-zero vector under powers of E and F spans V37 this
representation is irreducible.

Remark. In the sequel, it will be more convenient to work with
the universal enveloping algebra U(sl(2)). This is an algebra generated by symbols E , F , and H that are subject only to the relations E F - F E = 2H, HE - E H = E, and H F - F H = -F. The relations are motivated by the properties of the Lie bracket

denote the set of homogelleous poly-

+ s where j E {0,1/2,1,3/2,. . .). Note

that v1I2 is the fundamental representation, and V 0is the trivial

iir

10

REPRESENTATIONS l ( 2 ) ) OF U ( s
2. The vector v has weight j.
Then j E {0,2/2,1,3/2,. ..), and there is a unique linear map

~rof U ( s l ( 2 ) )is determined by assigning to E , F, and H operators on a vector space that are subiect t o the relations above. The enveloping algebra

. ,

-

L

--

--

-

-

- --

----e v is a vector in the

4 : V J+ W

such that $(x2"

= v and such that . 1, commutes with 1

representation V . In the discussion of Section 3 , the representations V J of d ( 2 ) and the enveloping algebra U ( s l ( 2 ) )will have quantum analogues. There is a quantum analogue of the group S L ( 2 ) , but we will not use it to describe the representations. should be indexed by

the action of U(sl(2)). (b) Every finite dimensional irreducible representation of U ( s l ( 2 ) )is isomorphic to V Jfor some j E {0,1/2,1,3/2,. . .).

Proof. Let vo = v , and for r > 0 let v, = F T v . We assume by induction that v, has weight ( j - r ) . Since
H v = jv and [ H ,F ] = - F , we have H V , + ~= H F v , = -Fur F H v , = ( j - ( r l ) ) ~ , + ~ ; vT+l has weight ( j - ( r 1 ) ) . thus Furthermore, we inductively assume that there are constants y, such that Ev, = y,v,-1. By using the relation [ E , F ] = 2 H , we have that E V , + ~= E F v , = 2HvT F E U , = ( 2 ( j - r ) Y T ) ~ T .

8
e3,m = 5J+m Y 3-m.
The first subscript of e is the highest weight of the representatloll <111dind~t f t 5 t h t tl~rnel~slon thc rcprt~.;crltdf a of ioll 5pr7(t\
(dl111jl
I )

+

+

+

+

+

Hence ?,+I = 2 ( j - r ) + y T ; since yo = 0 , we have y = r ( 2 j - r + 1). , 0 fol , o n ~ cI I,c~c,l~l\c.f * I ? f n r t e t l ~ m r . n s i o ~and \ I ~,~ Yo\\, ,

= LJ $ 1) wlllle the sccond ~~ldlcates icelght of the tile

vector. Note that j and m are both half-integers and that j and j - m are integers. In this notation,

+m

vo, v l , . . . are eigenvectors for distlnct elgenvalues of H. Suppose
that v, = 0 and v,-1

# 0.

Then E v , = 0 = r ( 2 j - r

+ l)v,-1.

So

Ee3,m = ( j - m)eJ,m+l, Fej,m = ( j
and

j = ( r - 1 ) / 2 E {0,1/2,1,3/2,. . .), the subspace of W generated by vo is spanned by the linearly independent vectors vo, . . .,v,-1,
and this subspace is isomorphic t o V J .This proves (a). (b) Let W denote a finite dimensional irreducible represerttation of U(s1(2)), and let w be a non-zero eigenvector of H . Let the integer r be such that E'w

+ m)eJ,m-17

Hel,m = meJ,,. (a) Let W denote a finite dimensional representation space for the algebra U ( s l ( 2 ) ) . Let E W denote a non-zero vector that satisfies:
1. E v = 0.

# 0 while E'+'w

= 0. Then v = E'w

satisfies the hypotheses of (a). Hence W

= V J where H v = jv.

2.2.3 Well Known Lemma.

2.2.4 Theorem. Every finite dimensional representation space

for U(sZ(2))decomposes as a direct sum of irreducible representations.

I

Proof. See [33]or [8]for example.

2.3 Diagrammatics of U ( s l ( 2 ) )invariant maps.

The Pen-

rose spin networks facilitate the computation of U ( s l ( 2 ) )invariant maps via diagrammatic techniques. At the heart of the networks are three elementary maps U,

n,

and

/

that are defined in Sec-

tion 2.3.1. Their relations are described in Lemma 2.3.2. The
111

3.

i
P

networks or spin-nets will consist of trivalent graphs embedded in the plane with non-negative half-integer labels on the edges. These labels will satisfy an admissibility condition at each vertex that will be made explicit as we continue the discussion.
2.3.1 Definition.

(1 fg U ) : v1I2 v1I2 -+ under the identification of C ,g, v1I2 y1l2 C = v 1 I 2 . = fg
(I fg n ) 0 (U 8 1) = ( = ( n 81)
0

I

,lit

170 + ~ 1 1 ,g, ~ 1 1 2 , 2

,

Consider the U ( s I ( 2 ) )invariant maps U :
: v 1 1 2 fg ~ 1 1 + 2

4.

( n 8 1) 0 ( I

fg

1) = ( I @ n ) 0 (1@
/

1)

: (v1/2)@3
1 18 2

v1/2

vO,

: v 1 / 2 fg ~ 1 1 4 2

where as before we identify C 8 '[fl/2 = ~

c = 1/1/2.

I

i

l f 1/2fg171/2, and I : v1I2 -+

v1I2 are defined on basis elements that

/

I

(and extended linearly) via

X(afgb)=b@a

for a , b € { x , y }

where i = G.Finally

6.

flou:C-+C

is multiplication by -2.
Proof. Items (I), (2), ( 3 ) and (6) are elementary computations.

] ( a )= a for a E { x , y ) .
2.3.2 Lemma (Penrose [27]).

Item (4) is a general property that holds for any bilinear. form n; similarly, item (5) follows for any "co-bilinear" form U .
2.3.3 Remarks.
v1I2

1. The maps U,

n, X,

and ( commute with the action of U ( s l ( 2 ) ) .

2. The fundamental binor identity holds:

Penrose and I<auffman introduced these

maps in a diagrammatic context. The domain of a map repre=I
fg

1 + (U o n) : v1J2 v 1 I 2 -+ 8

8~

1 1 : 2

sented by such a diagram appears at the bottom of the diagram,

16

'THE CLASSICAL D Q U A N T U~M- S Y M B O L , S AN ~

I

and
I
(1

The key observation about the diagram algebra is that ~ l a l l a f isotopies of arcs (that are properly and disjointly embedded ill a rectangle) are generated by the topological moves that corre-

.-z

-

i

hk

H

.. ,-

I -nL+. . I . A

( ...
I

21-L-1

spend t o (a) the relationship depicted in Lemma 2.3.2 (3) and (b) example, in the illustla: . , r c ~ l c l l d n ~disfg i l l ant critical l>ointq.
tloll above the Criticdl p01111~~ e l ~ l e ~ e l ~1 1lC d l l t1l l g
1 1 ~) ~ l ~ i l (I0'' ( / " ] ~

and those representing h3 can be pushed up. Algebraically, this The diagra7n algebra consists of formal linear combinations of certain diagrams. The diagrams are generated by the diagralns representillg I alld hk for k = 1,. . ., 2 j - 1 that are indicated above. two such diagrams can be juxtaposed vertically t o represent the product of two of the elements. Having been so juxtaposed, the product is rescaled vertically to fit into a standard size rectangle. Two di%pms that are isotopic via an isotopy that keeps interchange represents the identity h1h3 = h3hl. The correspondence between the Temperley-Lieb algebra and the algebra of diagrams shows that the dimension of T L n is the nth Catalan number,

A(:),

where n = 2 j . This result follows

by establishing a one-to-one correspondence between the possible diagrams and the collection of legitimate arrangements of n pairs of parentheses (See also [16]). Next we let the ground ring R denote ihe complex numbers, and choose matrix representations for the symbols U and n.
2 4 1 Lemma. Let j be a fixed element in {0,1/2,1,3/2,. . ..

the top and the bottom of the diagrams pointwise fixed represent

the same element in the diagram algebra. For example, the product
h2hlh3

is depicted below. Always the element on the left of

an ex~ressiollis a t the top of a diagram; thus the bottom most

18
I

THE CLASSICAL QUANTUM j - s y ~ ~ o ~ ~ AND 6

REPRESENTATIONS OF

U(sz(2))

19

functions (~1/2)@2.1 (~1/2)@2.1 , The image O(I) is the identity, and for k = 1,.. . , 2 j - 1, the generator hk is mapped to the composition Uon where these are acting on the (k, k

roof. The proof depends on standard facts about the represenation theory of GL(2) and C,. We refer the reader to the excellent text, [31] for details. Let n = 2j. Since dim (TLn) = Catalan(n) we must prove that dim (B(TLn)) = ( T ) / ( n

+ 1)- The binor iden-

+ 1) factors of the tensor product as in Section 2.3.1.

tity shows that B(TL,) = b(C[Cn]), where @ is the linear extension of the representation defined in the remark immediately above. We will establish that dim (@ (C[xnI)) = (',") /(n + 1). The representation

il
I

I

Proof. It is necessary to check that O respects the defining relations (1-5) of the Temperley-Lieb algebra. The calculatioll follows from the diagrammatics of Lemma 2.3.2 part (3). J I As we continue to discuss U(s1(2)), we will work with the Temperley-Lieb algebra under this representation witllout explic-

6 is decomposed

as in [33]. Namely, as En repre-

I'

(v'l2)@" = $ T ( m ) d ( T )

llill
I

itly mentioning the map 8. Our justification for this notational abuse is given in Theorem 2.4.3.
2.4.2 Lemma.
i 5

(1) the index T ranges over all 2-row Young frames with n (i.e. T = (r,s), n = r

Let 6 = -2.

11 ~ J ( ~ ~ I ~ o I I J O ~ 1 ~ ) / I (

f i r a n y j E {0,1/2,1,3/2,. .I, I ~ ~ ~ 1~, , , , l / c , / , o , j q J , J y2! ,,,, 2 ,
,

+ s, and 0 L r I s);
lllclt

,,,,

(2)t h e s u l 1 3 $ a ~ d \ITrr is t h e irrc~I,,cil,lc rel)resentatioll of
corresponds to the toung lldllle 1 ;

letters zlzto TL23 that 2 yzven by p(ak) = I + hk where ak zs the s transposition that interchanges and k + 1.

(3) the exponent d(T) is a positive integer that, incidentally, is equal to the di~nension the representation of GL(2) correspondof ing t o the Young frame T. jyow c [ x n ] is a semi-simple algebra because the group En is finite. The Wedderburn theory of semi-simple algebras [g], applied to C[Cn], asserts that as algebras

,

Proof. Clearly, the images (under p) of distant transpositions commute. Furthermore,
p(ukokklok) = I + h i

/I1

l1

+ h x t ~ hihi51 + hkil hk = p(okklokok*l 1. +
+ 6hk = 1.0
-

Finally, ( I t hk) 0 (Ithk) = 1 $ 2hk -a factor of the representation

Remark. The binor identity shows that the 1~omomorphism is p
: CZ3-+ A U ~ ( ( V ~ / ~ ) @ ~ ~ ) where the

-

@(C[Cn]) $ ~ M a t ( nX~n ~ ) % square

permutation group acts on ( V ~ / ~ ) by permuting tensor factors. @~J
2.4.3 Theorem. In case 6 = -2, the representation, B of TLz3

where nT = dim WT and Mat(- x -) denotes the algebra matrices. It follows that

on ( V ~ / ~ ) @ ~ J , is faithful for evert, i E -!0.1/2.1.3/?

1

I

dim ( j( C [ C , ] ) )=

n;.

For T = (r, n - r ) with 0

5 r _< Ln/2] (where 1.1 denotes the

A filling of the frame (s, n - s ) with the integers 1,. . .,n tother with a filling of its mirror image with the integers n $

greatest integer function), we have

.. .,2n yields a filling of the rectangular
m on the right is no larger than G(n, n).
This is the number of ways of filling in the n boxes in the Young frame T with the integers 1 , 2 , . . .,n in such a way that numbers increase across both rows and increase down all columns. The Young frame has n - r boxes on the top row and r boxes on the bottom row. The proof will follow from the following interesting combinatorial identity for Catalan numbers:

frame. Therefore, the

On the other hand, let a filling of the rectangular frame be iven. Then consider the subset of the rectangular array that ontains the numbers 1,.. .,n. This subset is convex and forms a smaller frame of type (s,n - s). Thus we have a filling of it and a filling of its mirror image. Therefore, G(n, n) is no larger than the sum on the right. This proves the combinatorial identity. Consequently, the representation is faithful.
2.5 Tensor products of irreducible representations.

Re-

Let G(r, n - r ) = n(,,,-,)
n-

denote the number of legitimate
I,

filling.; of the Young frame nxilh
1.

I ~ n ~ c > if 1 nri

1 1101 l o r ? ] J.(IU. ;I ~

nd

boxes on the top. For a two ron. rectangular array,

so in fact G(r, r ) is the rth Catalan number. We wish to show that

I
.

call that if V and W are spaces on which the group SL(2) acts, then there is an action given on the tensor product by g ( v @ t u ) =
911

~ I L ) . .\1i

element

. in the Y

assoriatctl 1,ic alge111.a. d ( 2 ) . a c t s
I ,

on tensor protluc~s the Leibniz rule, . Y ( u via

L ru) = St?) i $ I u v @ X(w) since the action is determined by differentiation. Notice that if v and w are weight vectors, then so is v @ w, and its weight

is the sum of the weights of v and w. Recall that V3 is isomorphic to a sub-representation of the 2j-fold tensor product of the fundamental representation via the map

,

(P3 : X I . ' ' ' ' x2j
Each term in the sum on the right is the square of the number of ways of W g g in a smaller Young frame. For s [n/2j, we consider the Young frame (n, n) to be decomposed as the union

++

1 -p

C

xu(1) @

. ' @ Xu(2j)

of a frame (s, n - s ) and its mirror image. For example,
2.5.1 The projectors.

The projection of (V1/')@'j onto the

image (Pj(Vj) can be written in terms of the Temperley-Lieb ele-

ments as the map

Observe that

+2j 0 + 2 j

= 4 2 , So that this map is indeed a projec-

2.5.3 Lemma.

proof. The proof follows by induction. 2.5.4 Definition. Suppose that a , b E { 0 , 1 / 2 , 1 , 3 / 2 , .

Let

j E {a

+ b,a + b - 1 , . . . , l a - bl

+ 1,la - bl). Such a triple of

half-integers ( a , b, j ) is said to be admissible. Notice that admissibility is a symmetric condition in a , b, and j . Define an U ( s l ( 2 ) ) In 3-5, the quantum analogues of these projectors are deinvariant map
:(~'/2)@23

-

(~1/2)@2a ( ~ 1 / 2 ) @ ~ ~

2.5.2 Definition.

Let

u= u : c

I

defined

C

+

( v " ~ ) @ ~ ( " -define ~),

~ 1 1 2 ~ 1 1 2 . H~~~~~ a 5 to be the composition

where , 1

is the identity map on the m-fold tensor power of v 1 l 2 -

The map ( p a @ p a ) o

Y

ab

0q5~-where
3

de is the isomorphism of

v with the symmetric tensors while pe(xl@. . - @ ~ 2 e ) 5 1 - . . ' 'x2e e = for 1 = j , a , or b - is called the Clebsch-Gordan map: V 3
+

The map fi is defined d u d y . Let C ( 1 , . . .,n}. Define

2.5.5 Theorem.

There is a direct sum decomposition

24

T H E L A S S I C AALN D Q U A N T U M j - S Y M B O L S C 6
Example. Consider v

where the sum is taken over all j such that ( a , b, j ) is admissible. Furthermore, i f ( a ,b, j ) is an admissible triple, then any U ( s l ( 2 ) ) invariant map V j

~ / ~ @ v ~ / t ~Theorem 2.5.5, According o .

v

+

V a @ V b is a scalar multiple of pa @ pb

o

this tensor product decomposes as the direct sum of V 0 and V 1 . 112,112 coincides with U while The map
0

0

q5j.

Finally,

and finally

y
Proof. The map

112,112 I (~xY)=~I~(x@Y+YBx).

y :"
,'

2.5.6 The Clebsch-Gordan coefficients. Let ej,t denote the

is U ( s l ( 2 ) )invariant because it is the

weight vector xj+tyj-t in

~j

of weight t . We have maps pa @ pb

o

composition of l T ( s / ( 2 ) )invariant maps for (Pa @ 1 ..

(

(4. ( x ' J ) ) )

/

and

G. TIle

(

V

ab

o

$; : vj -+ V a 4

vbprovided that ( a ,b, j ) are admissible.

follnu~~l

follows by cornputatiol~using

in the sum
8 pb

Lemma 2.5.3; thus

y

ab

#
3

0 for ( a , 6, j ) admissible. The tensor

product V a €3 vbhas dimension (20 of V 3has dimension ( 2 j 1). Since

+

+ 1)(2b+ 1 ) while the image

(y

3

(A (e3,t)l) = u+v=~

c:;t:ea,U 8 eblv.

2.5 -7 Lemma, The Clebsch-Gordan coeficients satisfy the fol-

lowing recursion relation

x(2j
3

+ 1 ) = (2a + 1)(2b+ 1 )
( j+t

+ l)ct:% ( a + U + l)~:;b:,v,t+l t (b + v + l)'::b:l,t+~. =

P isomorphic to

(

( 3

)
v'.

( j t t ) ! ( j- t ) ! ( a + b - j ) !
is the only s p a c e of

v

8

v

is

V3

--+

Consequently every U ( s l ( 2 ) ) invariant V a 8 V bmust be a multiple of this map.

Z,w:~+w=3-t

( a u ~ ) ! ( b v + w)! z!w!(a - u - z ) ! ( b- v - w ) ! '

+ +

+

1

26
11

THE CLASSICAL QUANTUM- S Y M B O L S AND 6j

REPRESENTATIONS OF

U(~1(2))

27

Proof. The recursion relation is found by applying F to both
sides of the equation that defines the Clebsch-Gordan coefficient. The closed form is determined by solving the recursion using the

By definition of 4, the weight vector $ t ( e j , t ) is the image under -kzj of

2.5.8 D i a g r a m m a t i c s for weight vectors.

I
a+u a - u

m

n

where m = j

+ t, n = j - t.
). Then the neight
1

a

~

6

1

- bv- 3

We represent x (resp. y) by a white (resp. black) vertex with a string coming out from the top: 6 (resp.

Y=----T
j+t

j-t

I

I

E

These conventions have been known to physicists (see 1161). It is convenient to introduce similar diagrams for dual vectors. Consider the dual vector space (v1l2)*. We represent the dual basis vectors x* and y* of tkis dual space diagrammatically by From the closed formula given in Lemma 2.5.7 for the ClebschGordan coefficients, one cannot easily see the symmetry properties
"P<

11.

7

aid

7

, respectively.

For parallel strings representing tensor

of the coefficients under replacing

ej,m

with ej,-,.

However, this

products of the fundamental representation, putting one of these dots on the top of the strings algebraically means that we take the values of the pairing among vectors and dual vectors. In particular, one. We have, then, that the dual to

symmetry and the probabilistic nature of the coefficients is more apparent in the network evaluation. Here we consider the space of U(sl(2))

X means the pairing between x and x* which is equal to

2.6 T h e 6j-symbols.

invariant maps

vk-+ Va @I V b@I VC.We will construct such maps

THE CLASSICAL D QUANTUM AN SYMBOL^
in two different ways. First, consider the composition
\
\
N

REPRESENTATIONS O F U(SZ(~))

(a, b, n), and (n, e , k) form admissible triples, form bases for the vector space of U(sZ(2)) invariant linear maps vk-+ va@vb @VC. fiere pabc = pa @ pb @ pc is the tensor product of the multiplication maps and d3 : ~j -+ (v1I2)B2j sends a a homogeneous polynomial in x and y to a symmetric tensor.

(

I2a

€3

','

--

f

Ii1 lIi 'I / I
11 /I

k
(v1I2)@2k ( v 112
~ 2@ a

(

112)8 2 6 @

(~.l/~)@2c

Proof. The triple tensor product Va @ vb @ VC decomposes as ($,Vn) €3 VC = $,(Vn €3 VC) where the direct sum
is taken over all n such that (a, b, n) is admissible, by Theorem 2.5.5. For each such n, (Vn @ Vc) contains at most one copy of vk.Thus h ~ m ~ ( , ~ ( ~ V ) ( vv b, ) a @ ~@VC) decomposes as a direct sum of the 1-dimensional spaces h ~ m ~ ( , ~ ( ~ vn ( vVC). Sim) ) €3 ~ , ilarly, it decomposes as a direct sum of the 1-dimensional spaces

I

for various values of j. Second, consider the composition

II~I
11
If

i t /I

y.'
I
I.

ab

=(

y

812C)o

y

nc

k

h ~ m u ( ~ i ( z ) ) ( V ~ , V3). Va €3
2.6.2 Definition.

I

&

for various values of n. The values of j and n are restricted so that (b, c, j ) , (a, j , k),
' ' (a, b, n), and (n, c, k) all form admissible Lrlples. Alternatively, if one of these triples is not admissible, then we may declare the

{

]

Define the Gj-symbol to be the coefficient

in the following equation.

corresponding map

V

to be the zero man. (Recall t h a t if f n ,-----" & I. A "

h ;\

is admissible, then so are the triples obtained by permuting a , b,

I

Pa @ Pb @ PC'

2.6.1 Lemma. The sets

a

V

b

A

)

a\

b\/c)

By convention, as the indices j and n range in such a way that (6, c, j ) , (a, j,k),

{

a b n

}

= 0 if any of the triples (0,c,j),

( a ,j , k), (a, 0, n ) , j n , c, k ) is not admissible.

/
I/

30

1

In the 'paces

hom~(~l(2)) (Vk7 Va

@ Vb @ Vc), we have the two

bases that are defined by these trees, and the ~ i - ~ is ~ ~ b ~ l the

change Of basis matrix. For example, consider the case when = = = = l/2- One can compute direct]y from the definitions that the possible values for j and n are 0 and 1, and that:

and

{ :;::;; ] -lP, { :;:;;: } { t :;:P } i ; { ;; ;; } ;:
= = 1, = 3/47
= 1/2.
A IYIII 5i1 o

Recall that a spin-net is an embedding in the plane of a graph with edges labeled by non-negative half-integers in which each vertex has valence The three edges coincident at a vertex must 3.

form an admissible triple. The half-integer labels represent the
spin carried by an edge. When we need to emphasize the number
of strings represented by an edge (and hence the number of tensor

factors of the fundamental representation carried by an edge), we will label the edges with natural numbers that are twice the halfsuffers from this minor inconsistency, but integers. The

J i ' S('rfioil

2 fi u r yivc ~

:,(t

I,(,(!

(

c,,,

,

we have found that the meaning of the labels is clear within the
coll~c'xt111
~til:t

h

!i I- \'

I1l1~'ll

-

addition, we assume that the embedding is in general Powe need t~ define an u ( s z ( ~ ) )invariant map (V112)@2a (v1'2)@2b @ as follows:
j
+

sition with respect t o a fixed height function: Thus each vertex
appears at a distinct level, the critical points on each edge are

(v112)B2J,for admissible triples ( a ,b, j )

non-degenerate, and these critical points are at distinct levels from
0

n ,lI
I

= +2J
ab

( /a+j-b 8 a+b--j8 /j+b-a ) n
3
0

( 2 4 ,
+2b)

.

the vertices. Furthermore, some edges may be marked with 'ymmetrizers:

+.

More precisely, we include valence 2 vertices in

The compositi~nirj 0
U(sz(2)) invariant.
2.6.3
ab

(+a 8 mb) : va 8

vb , is also vj

which the two incoming edges have the same hbel. ~h~ principal results of the current section are the orthogo-

computations. ~h~ elegance of the spin-net notation that we have developed so far is it facilitates Otherwise tedious calculations. There is slight disadvantage in that the calculations are performed in the tensor power,

nality and the EEott-Biedenharn identities that are satisfied the ~ j - ~ ~ We will~give ~ . ~ b l proofs of these relationships (and others) that are simply manipulations of diagrams. To this end we state a diagrammatic lemma.

32
2.6.4 Lemma.

THE CLASSICAL QUANTUM AND 6j-SYMBOLS
The following relationships hold among the

U(sZ(2)) invariant maps

)\ ,

y , 6,

Inl

and

fi.

(Here we iden-

tify C 8 V and V for any U ( s l ( 2 ) )space V . )
1.
(In

8 ; ) 0 ( 6 8 l , ) = ~ n = ( f i ~l n ) o ( l n 8 6 ) .

Proof. Part 1 follows from Lemma 2.3.2 part 3, and induction
since the cancellation of a U with a

n can occur regardless of the

tensor factors on which those maps are acting. Part 2 follows by induction and part 5 of Lemma 2.3.2. Part 3 follows similarly using part 4 of Lernma 2.3.2. Part 4 and part 5 are proved using

2.6.5 Remark.

Consider the collection of proper embeddings

of two and three valent graphs in a rectangle whose edges are parallel to the coordinate axes in the plane. The free end points of the edges of the graph are embedded in the top and bottom edges of the rectangle. If two such embedded graphs are isotopic via an isotopy that keeps the boundary fixed, then there is an isotopy between them that can be decomposed as a sequence of moves that are the diagrammatic descriptions of items 1 through 5 in Lemma 2.6.4. The valence two vertices are represented in the Lemma by the projectors, and the valence three vertices are represented by the Clebsh-Gordan maps. To find such a nice isotopy, one replaces a given isotopy by one that is in general position with respect to the height fuliction defined on the rectangle. The existence of the generic isotopy is guaranteed by a transversality argument, and a similar transver, , i ~ , d ~

~h~~ we have the following calculation.

C I I I ~ I ( \ I( I ~~

(

~

(

~ : I ~, > ~ I

~

tl , O ~ o l j \~ I ~ I ~ ~ ~ ~ I~ I ! . I I ~ ~ : ~ I I ~

-

1

,,

of pieces each of which is of the diagrammatic form specified.
2.6.6 Theorem (Orthogonality).

Suppose that (a, b, n),

(c, k,n), (a, b, m), and (c, k, m) are admissible triples. Then then

6j-symbols satisfy the following relation:

Proof. Define

Therefore,

THE CLASSICALD QUANTUM AN SYMBOLS

REPRESENTATIONS( s z ( 2 ) ) OF U

39

. This complex is depicted in Figure 1. Each face
of the Zdimensional complex arises as the cartesian product of an edge of the tree and a unit interval; the interval factor is thought of as a time parameter in the deformation between the two trees. The vertex of the complex occurs as the edge labeled b passes through the lower junction of three edges. The 2-dimensional complex has

I vertex, 4 edges, and 6 faces; thus it is the dual complex to a
tetrahedron as indicated in the figure. Let us associate t o this complex a 6j-symbol. Then consider the orthogonality conditions and the Elliott-Biedenharn identities. Either side of each equation can be similarly thought of as a 2t l l l l l e ~ l ~ l o ~cell ldl

co~iil)lcx.1'01

C ~ ~ I I I 111 ~ ' I ~ L I I C ' ~ I C ,

2 ulle

\icil: O [

t11e

orthogonality relation is depicted both as a movie description and as a 2-dimensional complex; the Zcomplex is the "time elapsed" version of the accompanying movie. The sucessive stills of the movie differ by a 6j-symbol and these symbols are associated t o the vertices of the 2-dimensional complex on the right of the figure. The trees in the stills are the trees that represent the maps in
2.6.8 Associating t h e 6j-symbol t o t h e d u a l skeleton of

the orthogonality relation. Figure 3 depicts the other side of the orthogonality relation, and the 2-dimensional complex here differs from the previous one by one of the Matveev moves [25, 16, 321. Simaarly, Figure 4 depicts in a movie fashion the three trees that cessive stills in the movie differ by a 6j-symbol. In Figure 5 the stills in the movie are the trees on the other side of the Elliott-

that occurs when the branch labeled by b'is moved from the

right side of the tree

to the left side of the tree

Biedenharn identity. The 2-complex in either figure represents the

THECLASSICAL QUANTUM AND 6j-SYMBOLS

Figure 2: A movie of Gj-symbols and ort.hogonality (left hancl side)

Figure 1: A movie of a 6j-symbol and the Matveev complex Figure 3: A movie of 6j-symbols and orthogonality (right hand side)

dual to the union of tetrahedra - two tetrahedra glued along a face for Figure 4 and three tetrahedra glued along an edge for Figure 5. These two complexes are related by a deformation that is also one of the Matveev moves [25, 16, 321. Thus the identities expressed among the 6j-symbols are diagrammatically expressed as deformations between 2-dimensional cell complexes. So the use of 6j-symbols in the construction of the Turaev-Viro invariant appears quite natural in the context of the representations and their diagrammatic realizations. However, there are two obstacles to overcome before the definition of the Turaev-Viro invariant can be made. First, the invariant Figure 4: A movie of 6j-symbols and the Elliott-Biedenharn identity (left hand side) is a sum over representations, and since we have irreducible representations for all non-negative half-integers j, such a sum would be infinite, so we cannot use the classical theory of U(sl(2)) to obtain an invariant. Second, the Gj-symbols that we have defined
( I J J 1101

r1111 ( ~ ~ ~ ~ I I , (l l ~l l lC~ tI ~ ~ IIII(I I ~ l ~ ) I

1 IIII, t ill,\

(

, 1 1 1 ~ 1 0 ~I ) ( >

associated to tetrahedra in any meaningful way. We will overcome the first obstacle in Section 4, by passing to the representations of U,(s1(2)) for q a root of unity. Professor Biedenharn informs us that this is the physicist's notion of renormalization since we are converting an infinite sum to a finite sum. We will overcome the second obstacle by normalizing the 6j-symbols and by showing that the normalized versions possess the desired symmetr'y while still satisfying orthogonality and Elliott-Biedenharn relations.

We now express the 6+symbols in terms of evaluations of certain spin-nets. In particular, we determine
2.7 Computations.
Figure 5: A movie of 6j-symbols and the Elliott-Biedenharn identity (right hand side) some of the symmetry properties of the 6j-symbols in the current normalization, and we find a normalization that has full tetrahedral symmetry.

44

THE CLASSICAL D AN

Q U A N T U M 63-SYMBOLS

KEYKESLN'l.A'l.lVNS O F U

(S1(L))

45

2.7.1 Topological invariance. Let a spin-net be given. This

Proof. This follows from the remarks 2.6.5.
2.7.3 Lemma.

network is a graph with its edges labeled by non-negative halfintegers that has only 2-valent and 3-valent vertices. Two edges incident at a 2-valent vertex must have the same label, and if edges with labels a, b, and j are incident at a vertex, then the triple (a, b, j) is admissible. Choose an embedding of the spin-net into a rectangle such that the endpoints of free edges (if any) are on the top and bottom of the rectangle. Suppose that the labels of the edges that appear on the top are a l , . . .,a,, and those that appear on the bottom are bl, . . .,b, plane. The embedding of the spin-net allows us to define a map where these labels are read
.

The value of

fi: ( v ~ / ~ ) @ "( v ~ / ~ ) @C~ is @
i

given by

if in(- l)#{"'"k'~)

~ j = f j forsome

j = 1, ...n,
j = 1, . . . , n

if {xj, fi} = {x, y} for all

1

where x j , q E {x, y} for all j = 1,. . . ,n.

from left to right once the rectangle has been embedded in the

Proof. This follows by induction, let us exemplify the formula.

In case n = 2, the non-zero values of n are as follows:

2

jectors and

are associated to the 2-valent vertices and maps are associated to the 3-valent vertices. At the bottom of
-

)\

the rectangle the v b s are mapped via q5 into the tensor powers of the fundamental representation, and at the top the tensor powers are projected back onto the Vas. Some care has to be taken with regard to the indices along the edges as we indicated in the definition given in Section 2.6.3. Finally, the embedding of the graph
'

2.7.4 Lemma.

For u

+ v = j , consider the map,pj o

(Pa @ q5b : Va @ vb- Vj. We have i

is to be in general position with respect to the height function on

2.7.2 Lemma.

The map

does not depend on the isotopy

Proof. Recall that e,,, = xa+" y
eb,, holds, and that the image under a symmetrized version of it.

class (rel. boundary) of the embedding of the given graph in the

, that a similar formula for 4 of a weight vector, ej,t, is

THE

CLASSICALD QUANTUM AN 6j-syMBOLS

REPRESEN

lAllUlV3

U p

u

\J1\"J/

proof. Since the given composition maps an irreducible representation V J into vk,it is O when k

# j.

And when k = j it is a

constant multiple of the identity. In Theorem 2.5.5 we computed

4a(xa+uya-u)4 4
=x
2 ~

b ( ~ ~3 + ~ € yb-')))

ia+b-~ia+b-~- I ) ~ - u (
(c,

u+u=1
( -])'-i'
-

+ 1, - ? ) ' ( ( L t 7 1 ) ' ( h t !))I
(2a)!(Lb)I

. -

Proof. Part ( 1 ) is a direct calculation that can be achieved by evaluating either side on the highest weight vector,

4J( 2 2 ~ ) . part

( 2 ) follows from ( 1 ) and Lemma 2.7.2 by rotating the vertex of the
right side of the diagram for ( 2 ) 180'. Parts (3) and (4) follow by on the tensor products of appropriate weight vectors.

+ 1 ( a + b - j ) ! ( a + j - b).( b + j

( ( a f b - j)!)2(a+b+j+l)!(a+j-b)!(b+j-a)! ( 2 ~ l ) ! ( at b - j ) !
- a)!(b

+j + a + I)!

2.7.6 Theorem.

(2a)!(2b)!(2j I ) ! t

~h~ next to the last equality is a combinatoric identity. The
proof that follows was indicated to us by Rhodes Peele. Consider

where

is a Kronecker 6 function, A, = ( - 1 ) 2 ~ ( 2 + 11, j
@ ( a ,b, k ) =

a + b +j

+ 1 ) such

that the value f ( a b

+j

-

b

+ 1 ) is

every element of f ( { a

+j

-

+ 2 , . - ., 2 j + 1 ) ) - We count the

elements of the set B in two ways-

50

THE CLASSICALN D QUANTUM A 6j-SYMBOLS
There are

('+,'tl) choices for the image f ({1, . .., 2 j + possible
+ 2,. . .,a + b + j + 1)) can be arranged +
+

I)), and each such choice can be arranged in (a j - b)!(b j -a)!
distinct ways while f ({2j in (a

v'f2involved). The loop closure of 'Czj gives the value Azj = (-1)~j(2j + 1) which is the
strings (= number of tensor factors of denominator of the right hand side. The value AZj gives the special case when a = 0 of Theorem 2.7.6. The closed network in the numerator on the right has the value @(a,b, j ) - this is why the function is named theta.

+ b - j)! distinct ways. Thus

f

+

Alternatively, B = UeBe where Be = { f E B : f ( a + j - b f 1) = a+b-j' I), so # B = #Be. There are (e-(a+j-b) ) possibilities for

2.7.7 Corollary. Let (a, b,j) denote an admissible triple of half-

xe

integers. Then we have equality between the following spin-nets.

the set f -'({l,. . .,!))

since the inverse image must exclude the

integers in the closed interval [a j - b integers in the closed interval [I, a

+ j - b]. Each such set can be arranged in l ! distinct ways. Furthermore, f -'({l+ 2,. . . ,a + b + j + 1)) is determined by f -l({l,. . .,!)) and this can be arranged in ( u + b + j - e)! distinct ways. So that

+

+ 1 , 2 j+ 11 and include the

Now let C = a

+ u, and let j = u + v, we have
2.7.8 Lemma. The 6 j symbols possess the following symmetry

(a t j - b)!(b

+ j - a)!(a + b - j)!.
Proof. Embed the spin-net
S

The required identity follows by rearranging the factors in the above equation. This completes the proof. The spin-net version of Theorem 2.7.6 is indicated in Corollary 2.7.7. Here we have labeled the spin-nets with the number of

in a rectangle with the edges

labeled m and p attached to the top edge while the edges labeled
s and t are attached to the bottom. By Lemma 2.7.2, the eval-

uation of the spin-net remains the same when the cross bar of

52

THE CLASSICAL QUANTUM AND 6j-SYMBOLS

the

mH
s

t

is rotated clockwise 180' while the end points of the

2.7.10 Lemma. Let TET(a, b, c, d,e, f ) denote the value of the

boundary remain fixed. Having performed such a rotation, recouple; the 6j-symbol
I

spin-net depicted below:

that appears is

, where

a sum is being taken over

u. Then rotate the (now vertkal) cross bar 180' counterclockwise.

We obtain,

1
Thus the coefficients are equal since pm @ p po a ba.sis for the set of U(sZ(2)) invariant maps V S V t i V m@I VP. @

Then,

,If

u

04,@I&

forms

Proof. The proof follows by recoupling and then applying Corol;). 1. . . 2.7. - and L C I I ~ L I ; ~ ( ~ f . [!(i]. I IS). .\ <IIJ I , 2.7.9
il<i:ti.l~

of tlic

Sagrammatic proof is shown below.

n
f

Proof. Any of these constants is the coefficient Z in the ecluation

That there is such a constant invariant maps by pm @ pp o

Zfollows since the space of U(sZ(2))
-

Y

vkP m
k

-+

V m@ VP is 1-dimensional and is spanned 4k. We leave it to the reader to draw the
e d c
a

o

corresponding diagrams.

54

THECLASSICAL AND

QUANTUM ~ ~ - S Y M B O T 9

1;
i
I

2.7.11 Remark. In [18] Kauffman and Lins give a closed for-

"-- gives that isoto] affect the spin-net evaluation. Lemma 2.7.5 indicates how the spin net evaluation changes when a crossing is introduced. Changing orientations can be achieved by twisting the trivalent vertices, and strings through as indicated in the diagrams below. by Here we have indicated crossing information So that the reader can see how to pull the diporam nn t h e r i ~ h t out to reverse the orientation of the projectec

[!
I
11

mula for the value TET(a,b,c,d,e, f ) in both the case at hand and in the quantum case. Furthermore, there are methods to cornPute the values of the 6j-symbols based on the Efiott-Biedenharn identity [2] or the recursive properties of the projectors [24].

rn

Section 2.8, we define four fundamental 6j-symbols, and use their
11

e i
i*
I'

values and the ELliott-Biedenharn identity to compute the values of the 6j-symbols in general.
,

2.7.12 Lemma. The symbol,

.. .

I

= T E T ( ~ , c, d, e, /I/ J I W , 6, I)W, 6,
8

e, m a , c,

~ ) o (c, e ,) ~ B

---

is invariant under all permutations of its columns and under the exchange of any pair of elements in the top row with the corresponding pair in the bottom row. Equivalently, the symbol

$1
I

[ 1 :]

kinvariant under the permutations of the set ({a, b, f >, c, d>, {b, c, e l , {d, e, f}}. {a,

(b+f-a)+(a-c-d)+(d+e-f

)+(c-~-~)

(I

(l1i
/I

Proof. This set -is the set of vertices of a tetrahedron with
edges hbeled by half-integers a, b, c, d, e, f , such that any element in the set forms an admissible triple. We choose an embedding of the 1-skeleton of this tetrahedron into a rectangle (see for example the diagram below). This embedded labeled graph is a spin-net and as such determines a U(sl(2)) invariant map

B~ Lemma 2.7.10, the normalized 6j coefficient

is the value of the embedded tetrahedral network divided by the factor J(@(a, b, f)O(d, e, f)@(a,c, d)@(b,c, which possesses e)l tetrahedral symmetry by definition. This completes the Poof.

[: r ]

I
I

c

+

c

(such

56

T H E L A S S I C AAL D Q U A N T U M j - S Y M B O L S C N 6

The orthogonality relation and the ElliottBiedenharn relation hold for the normalized 6j-coefficients i n the following form. Orthogonality,

2.7.13 Theorem.

.{'J
d

k

p } n

Elliott-Biedenharn:

Proof. These rrs~rltsfollorv from Tbrorerns 2.6.6 r~,,,rl 2.6.7 substitut,io~~, ck~,~iceliilg allti Os 2.7.14 Theorem.
OIL eitlier

side oi. tile equations.

= x n( - 1 ) 2 n ( 2 n + 1 d [ b f n ] [ ) j c

a n p d k j

]

~ L A S S I C A LA N D

QUANTUM 6j-syMBOLS

REPRESENTATIONS O F U(d(2))

Therefore, comparing the coefficients for fixed m and p,

P r o o f o f p a r t (6).

a

Next we observe that part (2) follows from (I),part (4) follows

recoupling using the 6j-coefficient

2.8 A recursion f o r m u l a for t h e 6j-symbols.

In

Biedenharn-Louck [2], a recursive method to compute the 6 j symbols is presented. We summarize this method here. First, we define the following four 6j-symbols t o be the funOn the other hand, damental 6j-symbols. Their values can be computed by means of

THE

CLASSICALD QUANTUMj - s y M B O L S AN 6

REPRESENTATIONS O F

U(sl(2))

65

the recursion relation for

+

given in Section 3.5 a t A = 1.

see[241

for details. An alternative computation in the classical case (the

case a t hand) is given in [ l a ] . sibility conditions force j = f f 112 and k = g f 1/27 yields

2s 112
2

112 x

+ 1/2
g+1/2 e-112 f+1/2

- 112

112 z

+ 1/2
g

+ 112

e - 112

f

- 112

2+

1 / 2 112 x - 1 / 2 g - 112 e - 112 f

+ 112

2-

112 112 x - 112 g - 112 e - 112 f - 112

(e+c-

j)(y+h-~+l)

This pushes everything back t o the case
(e- c

+ f ) ( e + c + f + l ) ( f + d - g)(g + h - + 1 )
f)(f
( 2 e ) ( 2 f)(% -d-g)

+ 1)(2f + 1)

(e+c-

(e-c+f)(e+c+ f +l)(fd+g)(f+d+g+l) ( 2 e ) ( 2 f) ( 2 f 1)

+

+

The coefficients A, B , C , and D are values (with suitable choices for j and k ) of the product

which can be
2.9 Remarks.

using the four fundamental 6j-symbolsIn the above discussion, we have reproduced

proofs of the important identities among the Clebsch-Gordan co-

I '

iR

/
I)

66

THE CLASSICAL QUANTUM AND SYMBOL^

efficients and the 6j-symbols by means of the spin network analysis of Penrose [27] and Kauffman (16, 181. The diagrammatic or graphical method seems t o be well known [3, 21 but only with the work of [19]and its sequels has this been brought to the fore-front of the theory. The identities 6.16-6.19 given in [19] for the representations of the quantum group are given above (for the classical case) as Theorems 2.6.7, 2.6.6, and parts of Theorem 2.7.14 (but not respectively). The fact that these proofs are expressible in terms of simple diagrammatic manipulations points to various levels of generalization. In the next section we explore one such level: knot theoretic quantization, in which the braid group replaces the permutation group, and a parameter is inserted into the binor identity. This notion of quantization coincides with the notion of a quantuln
Z ~ O I I < \ \I ~

3

Quantum

!!/:
F $1

In Set3.1 Some finite dimensional representations. tion 2 we showed how the tensor products of finite dimensional of sZ(2) can be decomposed. In this section we mimic this classical theory in the so-called quantum case where the representation spaces are spaces of hOmogelleous polynomials in two variables that only c ~ m m u t e P U to a parameter. ~t is these representations that give invariants of '$dimensional manifolds, and physical applications are found in statistical mechanics where they provide solutions to the YangBaxter equation. Recall we have ~ o u l ~ IuC U U L I I J ~ ~r;= y l = - r r u w v A u r r v I I

/ I/

/

ii
il
h

--

--\-I

--

V J = {llomogeneous polynomials of degree 2 in x and Y). Namely, . 7
l f (1

( I ( > \ c I < ~ j ) ( l ( \ I>\ . J I I I I I ) O

I ~ I ,

10

,-,,

<,It</

(,

, [, ,

+ I ) = 27
L;[L"~J")=/)I

, + I !>-I

for example [ll]for further references.) Some puzzles remain about these techniques. Combinatorial identities arise - for example, recursion relations among the Clebsch-Gordan coefficients. Are there simple discrete probabilistic reasons for these formulas, and if SO, how are the more elementary combinatorial formulas expressible diagrammatically? In the next section we turn to the finite dimensional representations of the quantum group Uq(s1(2)). will see the direct We analogy-with the classical case as the theory develops, and in Settion 4, we examine in detail the special case when the quantum parameter is a root of unity.
n. .

tJ

F ( x ~ Y = ) a-1 y b+l ~ ax
(a- b n ( x a y b ) = -xayb. ) 2

...:+L +Lo

nnt-ltinn

h o t t ~ ~ adanted t o

the theorv of weights,

el,m --

.-

z3+m

Y3-m

Eel,m = ( j - m)ej,m+l

-

Fe3,,,, = ( j

+ m)e3,m-1

Hel,m = me,,, from which it follows that

IE, F]ej,m = 2me3,m
67

Ill
/I

T H ECLASSICALD QUANTUM ~ AN ~

- S Y M B O ~ ~

We want to "quantize," which, in our context, means introduce a parameter. Roughly speaking, we want to replace integers n by

second, the action of E , F and K is given by EeJ,m = [ j - m]e~,m+l Fe3,m = [ j mIe3,m-1

20

[nlq = 9n-1
If q

+ qn-3 +

. . . + q-(n-l)

= qn - q-n q - q-l

+

'

# 0 is fixed, we will write [n]for [n],. By definition [0] = 0 and

Kel,m = ~ ~ ~ e 3 , m from which it follows that
[E, F]e3,m= [2mIe~,m-

[I] = 1. The theory of the quantum group U,(s1(2)) is particularly interesting in case the quantum parameter q is a root of unity because [n]= 0 when in Section 4.
i

q2n

= 1. This case will be covered in detail

~h~ representation V i is irreducible provided /I4' 1 for #

5

5

0,1, -1 E C . A representation of Uq(s1(2)) a vector space W with operators E, F , I< ( l i " = " o H ) is such that I{2 - I{-2 11 - , q2H - q-2H - , [E, Fl = 4 - 4-I Q - 4-I T I? = q E i

3.1.1 Definition. Let q

#

2j.

1

Prnnf .
A

Comnute d irectly the actions of E F , FE, K E , E l i , K F , ---..

and FIi. The identities for the l i e brackets follow by manipdating the rational functions of q that result. To prove irreducibility,
C .

. ^ "

_

,

__. 1:

r ,

liF = y

- l ~ ~ <

lilese[orc, tile illla,ge

ally

null-Lc.!.o

\cCli)l

~ l i ; i ~ ( 'Ii ) O \ ~ ( ' l ~

where [E,F = E F - FE. The quantum group Uq(s1(2)) the alI is gebra over C that is generated by E, F, I<, K-1; the elements and
i:

and F spans vi. o n e of our present goals (Theorem 3.4.1) is to give concrete realizations of the abstractly given representations V i . The realizations will be !generated from the fundamental representations

are subject to the relations specified above. For the tirne being, we will not deal with this algebra directly, but instead we work with it via its representations. ~ h u we will only consider s the entire algebra (and co-algebra) ea: post facto. 3 1 2 Theorem. . The representations vj -of s1(2) .. be quantized i n the following sense: Let q # -0,1, -1 E C . Let j {0,1/2,1,3/2,. . .), and let A E C , where A2 = q. ~h~~~ is
a ( 2 j l)-dimensional representation of ~ , ( ~ 1 ( 2given ab)) stractly as follows: First, a basis for V; is

v 1 l 2 by tensor products. We can take tensor products of quantized
representations because the algebra Uq(s1(2)) has a comultiplication. Rather than specifying the comultiplication, we wig specify the action of u q ( s l ( 2 ) )on tensor products of representations; from these formulas the comultiplication can be derived.
3.1.3 Theorem. Let U and V be representations of U q ( ~ l ( ~ ) ) -

can

+

~i

( A always, ~

#

o,l,-l).

Then there is a representation of

~ , ( ~ 1 ( 2on U 8 V given by the following formulas))

THE CLASSICAL D QUANTUM AN SYMBOL^ '1

71

lassically, the obvious representation of the symmetric group

Proof. Three relations must be verified.

0

2 j (permuting

factors) on ( ~ ' / ~ ) @ ~ j commutes with the action of

3.1.4 Remarks. The representations for the quantum case wiU

sz(2) on the same space. The subrepresentation on which EZj acts

be made explicitly analogous to the those of the classical case. In particular, we will find the quantum analogues of the spaces of homogeneous polynomials as irreducible representations and the identification of these with subrepresentations in the tensor product of copies of the fundamental representation for generic values
dji~i

1
,u

.d

trivially, namely the space of symmetric tensors, is isomorphic to the representation VJ of sl(2). Since representations of the braid quotient representations in the quantized situation, rather than subrepresentations. The Artin braid group B(n) is given by gen321 .. erators and relations as follows: (sl,s z,... s : sxs, = s , s t if l k - j l > l ;

1'

1;

of the Parameter 9. These subrepresentations are the images of projectors that are deformations of the projectors in the classical case. Via such identifications, we will be able to perforlll the dia~ , r a n i n i a l i c(. (
~ ~ i ~ p 1 1 1 ~ 11 i1 o ~ 1 1 I I C r l ~ l c ~ l ~ s ~ l l f j, l O , f , 1 r i I~ 10

1"'
lir

1jiC,
Sksk+lSk

sical case. Specifically, we will show in the quantum case how

11 1
I!'
I'

= ~ ~ + ~ s k s k + k = l " A - - - , n - 2, if l

to decompose tensor products of representations (Clebsch-Gordan Theory), and develop the 6j-symbols and verify their important

Remark

~h~ braid group is depicted graphically as indicated below. ~ ~ l t i ~ l i ~ ~ t i ~ nvertical juxtaposition braid is achieved by

the so-called trace 0 representations that arise. To construct the concrete realizations of the quantum representation spaces, we will find a natural quotient representation of (v;/')@~J that is U,(sl(2))-isomorphic to V i . To describe the quotient, we introduce an action of the braid group B(2j) on ( ~ ; ' ~ ) @ ~that 3 commutes with the action of U,(s1(2)). Furthermore, for generic values of q or for q an rth root of unity (where r
3.2 Representations of the braid groups.

> 2j), we will

The braid diagram for s l s z s l s ~ l

72

THE CLASSICALD QUANTUM AN SYMBOL^

b

Q U A N T U M sl(2)

3.2.2 Definitions. Let A

# 0 E C be fixed, and let

;
Proof. This follows by computation.

Here we redefine the matrix representation of the maps U and to apply to the quantum case. Define

n

3.2.4 Definition. We define a Uq(s1(2)) invariant map (called

a positive crossing)

nn = n : vii2@ viI2+ C = V i

via

as follows:
?/ @ x

I--+

-iAW1,

and

Y@Y++O.
Define U A = U : V j = C

-

v" ;

@

vlI2by the formula

Observe that

(
\

-1

A

Let the negative crossing be defined as follows:

)

= A-1

[ ] 4- A [ 1 1 ] .
@
/J

k

3.2.3 Lemma. Let A2 = q.

i
the domain is

L

A

.

The
2.

UA and

n~ are Uq(s1(2))invariant.
~

~ A O U A =O A : C - + c

in section 2, the symbol I denotes the identity map, but here

is multiplication by
3.
U A o nA =

- A-2 = -[21. 2
V/ ;2
-+

v:I2.
There is a representation r~ of B ( n ) o n

u : vl/2 n
(x@x)=

vy2,g,viI2

3.2.5 Theorem. ( v : / ~ ) @ ~ by defined

is given by
-.

U

.

n

n l-J

(Y@Y)=~,

rA(sk) = -/
k-1

I ...IY
n-k-1

and
U

4
n ( Y @ x ) =X
@

*+

f o r k = 1,2 ,..., n - 1.

y-q-ly@x.

Proof. It is clear that the images of distant braid generators commute. That the relations sksk+lsk = sk+lsksk+l hold in the representation for k = 1 , . . .,n - 2 is a direct computation wl

t

11

74

THE CLASSICAL D QUANTUM j AN 6

-

s

~

~

~ Q~ A N T U M d ( 2 ) U ~

75

li
iI

t

'
8

is usually performed diagrammatically and depends on the far that

proof. The situation is analogous to Lemma 2.4.2 and
rem 3.2.5. So details are omitted.
3.3.2 Remark. The relation

iJ
il

where

( n @ I O ( I @ OX(* )) = ( I @ n) 0 ( X ( r f ) €4 ( ) X(&) denotes the positive or negative crossing depicted

above. (See [16] for the diagrammatic version, or compare with Lemma 2.3.2.)

h i = (-A2 - ~ - ~ ) h k
in T L , ( - [ ~ ]cuts the braid algebra down to a finite dimensiona1 ) algebra.
3-3.3 Remark.

I
1"
(J

1

I

Let A E C , q = A2 # O,1, -1, The acttons of Uq(si(2)) and of B ( n ) (via T A ) on (v:/~)@" are commuting actions.

3.2.6 Theorem.

bili
,I; i

The formula for p ~ ( s k is referred to as )

.r
ji '; .:

Proof. This f~llows because T A is defined in terms of the ~ ~ ( ~ l ( 2 ) )
maps and

bracket identity. This identity encapsulates Kauffman's simplification of the Jones polynomial 1161. Recall that the bracket identity

1

4

1.
The repc.,,,.:;,ii,
via f h c .
illtcl.r!lc~c]i;il.i.

Il

1

3.3 A finite dimensional quotient of C [ B ( n ) ] .
l'p%'llf~t~ions

is given diagrammatically as:

Y..l

1~'Pl.p tliscovcrrt]

finite dimensiona.1 algebras that are quotiellts of tile (inlil,ite dimensiona.1) group algebra of B ( n ) .

'y
/
- ~ - 2 - ~2

=jL(

X )+
K , of

1 I).
by removing each

Given a knot diagram,

a knot, K , the bracket identity

I

I@ :& ,j;)

Next we consider the Temperley-Lieb algebra T L , ( ~ ) where

2

and where 6 = - ( A 2 f A-') for some A

#0

is used to compute a polynomial

(K)in A*'

E C . Recall

ig

that T L n is generated by elements I and hk for k = 1,.. .,n - 1 that are subject to the relations given in Section 2.4.
3.3.1 Theorem. For n

crossing via the bracket identity and associating the loop value to each of the simple closed curves that results in

any of the daughter diagrams. The bracket is an invariant of the
regular isotopy class of the diagram, but not of the knot type
TO

2 2 and A # 0 E C there is a surjective

A ' -

algebra homomorphism PA : C [ B ( n ) ] , TL,(-[2]) that is defined on generators as f o ~ l o ~ ~ :

obtain an invariant of K, define
--

L ( K ) ( A )= ( - A ) - ~ ~ ( ~ ) ( K )

..

76

THE CLA'SSICAL AND

Q U A N T U M 6j-SYMBOLS

QUANTUM

d(2) Let

It is clear that the representation r~ of the braid group defined in Theorem 3.2.5 factors through PA. The algebra TLn(-[2]) is represented on 2" v' I @ ( ) ; via the map

3.4 A model for the representations Vi.

be the tensor algebra of v;I2, and let LA c as in the classical case but with ing result:
3.3.4 Theorem. The representation

en(^;/^)

=
A

. We have the follow-

be the two-sided ideal generated by qx € y - y @ x (where q = A2, 3 as always), and let 112 ~ 2 A = LA n (VA )@ 2 j . j Then there is an isomorphism

is faithful. The case when A = 1 is covered in Theorem 2.4.3. The case i = -1 i t l l l c > ,<IlllC*<l, -4 = 1 l ~ ~ ~ < < lI )L lL~L<1~ f J , L ( - [ 2 ] )< l l l < l I t \ O I representation OA depend on A only through q = A2. The proof of the generic case and of the case that A is a 4rth root of unity and r

~ e n ( v : / ~ ) / ~ a C[x, Y]/(YX qxy) -

1

between the quotient algebra and the algebra of polynomials in tile I L ~ I I C ~ I ~ I I III ;~ I ,\ ., ,~ 1.11)1(~\ 1 r ~ i l ( l t/ \\ I I O I C 11 1 = (1 I I / (TI\(> I parameter q commutes with x and y as it Moreover, we have T ~ ~ ( v ; / ~ ) /= ~@ (v1l2)@2j/ ; L L
j=O,l,

IS

a member of the

> n is explained

ground field.) We will identify these two algebras in the sequel.

in Section 4.2. The remaining case is

proven in [7].
3.3.5 Notation. Let IA(n) denote the two-sided ideal in

...

TLn(-121) that is generated by the set {hl,. . .,hn-l}.
3.3.6 Lemma.

where (~1l")ej
%

/ L"Aj

The ideal IA(n) is a proper ideal, of C-vector

space codimension 1 in TL,'(-[2]).

can be identified with the space of homogeneous polynomials of degree 2 j in x and y, where yx = qxy.

Proof. The space TLn(-121) = TL, is spanned by 1 and the
monomials in the hk, so the codimension of IA(n) can only be 0 or
1. We need only prove that 1 6IA(n). This is easily seen from the

A direct computation shows that

~2 = I A ( ~ ~ ) (112)A V

82,.

representation of TLn(-[2]) on ( ~ 1 ' ~ ) ~ " . l ( x 8x 8 . - -8 = For x) x @ x @ . . . @ x but ,
X @ X

112 @23 Because IA j ) is an ideal in T L2, (- [2]) whose action on (VA ) (2 commutes with the action of Uq(s1(2)), it is clear that L? is a

@ - - . @ x I A ( ~ ) ( x @@ . . . @ x . @ x )

78

T H E L A S S I C AA N D Q U A N T U Mj - S Y M B O L S C L 6
3.4.2 Lemma.
1.
qXayb)

subrepresentation of ( ~ l / ~ )for thejactions of all three of T L Z j , @ ~

B ( 2 j ) , and Uq(s1(2)).
3.4.1 Theorem. The actions of Uq(s1(2)), ( 2 j ) , and T L z j on B

= [ a ] ~ a - b - l x a - l Y b+1

(v1I2 2 J /canybe described as follows: A 1@ ~
3.
1. For T L 2 j :

l i ( x a y b ) AaVbxayb =

IW = W ,
2. For B ( 2 j ) :
S ~ W =

hiw = O

for all w E

(vA~~)@~~/LZ

Proof. Notice that
and K y = A-lx.

v:/'

consists of the set of linear combinations

of x and y. Thus E x = F y = 0 , whi1e.F~ y, E y = x , K x = A x , = These computations form the initial steps of inductions that will follow.

A-lw

for all w E ( V , / ~ ) @ ~ ~ / L : :

@ ~ We compute, Ii(xBa @ yBb) = ( K x ) B a @ ( ~ y ) = Aax@" @ A-by@b = Aa-bxBa 8 yQb where, for example, xga is the tensor
product of a factors of x. Thus the third identity holds. @ By Item ( I ) ,we mean that F ( x g a @ yBb) [ a ] ~ ~ - ~ - ~ x @ ~ - '

3. For Uq(s1(2)):There is an isomorphism

-

, : 1 .;: - j 1 .. II / ' )'???.i,/ 2 i . .
,
1

4

: Y ~ ~ ~ ( + o d L,t). m~

First consider the case whe~e = 0. Illcll lllduct

011 (1.

I.0'

F ( x B a ) = F(x@~-'B x > Henceforth, we will identzfy V i with ( V : " ) @ ~ J / L ~ the isovia
=~
- 1 ~ @ 8 FX ~ - l

+ F(x@"-') 8 K x
y
(mod L A )

[ a ] ~ a - l ~ @ a - '@

Proof. Item ( 1 ) follows because ~ , ( ( V : ~ ~ ) @ ~ J Item ( 2 ) c L?. )
follows from (1) and the bracket identity. Item (3) follows from the next Lemma which indicates that in the case at hand E, F , and I i act as quantum differential operators; so these actions are analogous to those given in Section 2.2.

because A1-a

+ [a - l]Aa+l = [a]Aa-l.
F(x@" @ yBb) = F ( x @@ yBb-' @ Y ) ~

For general b 2 1,

-

8 Y @ ~ - ' )@ F y

+ F ( x B a €3 ymb-l) @ ICY
(mod LA.)

- [a]~a-bx@a-l
The

@

yb @ A-ly

of ( 2 ) follows along the same lines. This completes
0.

the proof of the Lemma

80

THE CLASSICAL D AN

QUANTUM

SYMBOLS

1
d
Observe that further that
Y

3 4 3 Remark. Let XA be the one dimensional character o ~ B ( ~ ) ..
such that xA(si) = A-'.
I 4

(If A = 1 or -1 then XA factors through
XA

+,Ais not defined when A4T# 1 for 1 5 r < n.

[n]! = 0; thus we assume

the ~ e r m u t a t i o n group En, giving the trivial character (A = 1)or

111'
41

the sign (A = -1). Thus

is a sort of "generalized sign".) Part

3.5.2 Theorem [13]. cf. [18, 23, 161. The element

4-1t TL,
1).

1:

satisfies the conditions:
1.

(2) of Theorem 3.4.1 asserts that the representation of B(2j) on the space of homogeneous polynomials of degree 2 j in x and y is through the character XA. In the next section, we will define for certain values of A, a Uq(sl(2)) invariant projector h, o(+" O/)O h,o(+! B I) = ( - l ) w h n o ( + k €3
=

'ijir

3
% -

2.
3.

$5 : (V;/~)@U

:

-+

(v~/~)@~J whose

I

4
"!

+ao

=

..

+; 04;
4 ;
=

4;.
for all u E I A ( ~ ) .

o

image is a subrepresentation that is isomorphic t o Vi. The projection is analogous t o the projection in the classical case, and it we will be able t o construct the quantum Clebsch-~ordan coefficients and the quantum 6j-symbols.
hc classica.1 ca.se, we will define idempotents, called the Jortes- 13Ve1ztzl
3 - 5 The Jones-Wentzl projectors.
111 analoRv wit], t

Moreover, any non-zem element in TL, safisfying these conditions must be equal to 4 ; .

proof. The recursion relation above defines the projector, 4:.
We will mimic the algebraic proof presented in [16], but here we
use diagrams. The diagral1lma,tic version of tlle recilr5ion relation

, ,

I

*

.

[11,-11,1,i

-,,

,

,L.4

(.,., I i

1

projectors, is depicted as follows:

I!

in the algebra of Uq(s1(2)) invariant transformations on ( v ; / ~ ) @ ~ .
3.5.1 Definition.

Let A

#

0,1, -1 denote a complex number, into

-3 9
5

and let A2 = q. There is a canonical embedding of TL,-l every generator in TL,-l.
-

TL, obtained by juxtaposing a straight string on the right of
rn

*
We will assunie, by induction, that
7M

We will consider al of the elements in . l

1

'j
'8 -

TL, t o also be in TL,+k for the rest of this section. Define the Jones- Wentzl Projector,

4 :

E TL, via the recursion relation:

+a =+A, €3 I + ln - 1 1 / [ ~ ] ( + 5€3~I )
where

that +i-lhi = hk+!-, = 0, for 1 being trivial. Then we show that

< k < n - 2 the case of +f

+;-l+al+ a l , and
=

h n - ~ (+fl

I)
by applying h, t o the recursion for 4; to obtain hn 0 (+A

4 :

= I = id :

vii2, y;l2.

O 1) =

83

ost horizontal line in the second icon from the left was absorbed e single vertical strings in the second and fourth icon into the ring to the left. In this way, the labels on the strings change om n - 2 to n - 1. (The labels on the vertical strings on the last on are omitted for type-setting reasons) Furthermore, we have composition of idempotents. So the above sum reduces to the

- (-PI

+ En - 11/Inl).

+ [n-l]/[n]

n-2

'I'HE CLASSICAL D Q U A N T U6 j - S Y AN M
w e obtain hn 0 (4: @ 1 ) by the identity

0

h n ( e 8 1 ) = -[n

+ l]/[n]hn(+f 11 8

and the projectors involving n - 1 strings are absorbed since these are idempotents by induction. So far we have shown that dempotents. Furthermore,

4 # ;

+;

are

0 because the coefficient of

1,

among quantum numbers that is easy to verify. Next we use this t o show that

n the sum is 1, as can be seen by induction. By induction and by this recursion relation, h k e = 0 = e h k if k = 1 , . . ., n - 2. Next we compute hn-l+i =

+f+;

= +A . n

The last sum is obtained again by the fact that (hn-1e-,)2 =
..

- [ n ] / [ n 1] - hn-l+i-l and the sum of these two terms is clearly 0. By the top/bottom symmetry of the diagrams, it follows that

' I ' , A ~ , - ~ 0. =

Thus we have an inductive construction for the

Next we show that these elements are unique. Any given element in the Temperley-Lieb algebra can be represented as al, +U where a is in the ground ring,

,1

denotes the identity, and U E

I A ( n ) so that U is a linear combination of products of the h j .
Suppose that g: = g, is a non-zero element of T L , such that

gnIA(n)= (0). Then g, = a ( ,+M = a21, + 2 o l +M2. In particu~
In the fourth term, the icon that represents (hn-l o 8 is replaced by the icon representing - [ n ] / [ n l ] h , ( + , ~ _ ~12), 8 lar, a2- a = 0 since In 6 IA(n).If a = 0, then 0 = gnu = g i = g,. Hence a = 1. Thus any non-zero idempotent f n that kills IA(n)

86
I !

THE CLASSICAL D AN

Q U A N T U M 6j-SYMBOLS

I

QUANTUM sZ(2)

87

can be written in the form fn = In+U1. Finally, gn = gn(ln+,gl) = (I, U) fn = f. This proves uniqueness 0 . ,

+

4
= .

First we recall the canonical inductive construction of the set of permutations. Let (k, . . .,n) denote the cyclic permutation of the elements k through n in C,; this cycle (k, . . .,n) can be written as a product of adjacent transpositions: (k, . . .,n) = (k, k
1, k+2)-. . -(n-1, n). To a bijection t : (1,.

For every permutation a E C,, define a braid 1 E B(n) as follows. Write a in any way as the product of 3 a minimal number of adjacent transpositions (i.e., transpositions
t

3.5.3 Notation.

+ 1)(k +

P 4

of the form

ak

= (k, k

+ I)), and let T ( a ) denote this minimal

t

I* 14
tr I

I

number. Then lift the product to B(n) by replacing each of the transpositions
ak

. . .,n), we associate a pair (k, tk ), where t(n) = k, t = (k, k + 1,. . . ,n)tk, and tk(n) = n so tk can ble regarded as a permutation of (1,. . .,n - 1).
-+ (1,

. .,n)

in the product by the corresponding braid gen-

In this way, the minimal numbf!r of adjacent transpositions that it takes to write t is 4equal to m e minimal number that it takes to ' ' " write tk plus n - k. Consider an ascellding path that starts from the lower left and travels upward through the triangle depicted below. ( 0 , l ) (1,2) ( 0 , l ) (1,2)
( 0 , l ) (1,2)

.

-

erator s k . The minimality of the product for a insures that the lift
b will depend only on the permutation a and not on the particular

1 4
I*'

'I

product representation chosen. Diagrammatically, the transposition ok is represented by n arcs running down the page in which the kth and (k cross. The corresponding braid generator
13v

+ 1)st arcs
+ 1st.
LI~IIS~OI-

(n-2,n-

1) ( n - 1,n) 1)

sk

can be represented
dlgcbld

...
. ..

(n-2,n-

t h e same arcs, where the kt11 arc crosses over t h e k
Next r e p ~ e s e n tthe b ~ a l dg ~ o u p I I ~ Othe I

01

mations on

(v:/~)@~ the bracket identity. via
- - [I].

Let

[XI

denote the

(071)

quantum integer that corresponds to the integer x. Finally, let [n]! = [n][n - 1 1

A permutation can be represented by such a path as product
of the switches to t the product of the E

3.5.4 Proposition [18]. The Jones-Wentzl projector is also

given by the formula:

P r o o f of Proposition. We must show that the formula above defines an element that kills each of the hk, 1 that the coefficient of

< k < n - 1, and

(0, I), (1,2), . . .,(n - 1,n). In fact, such descriptions of permutations by paths always use the minimal number of adjacent transpositions necessary to write the permutation. We compute the coefficient of 1, in the sum ~ u E z , , ( ~ - 3 ) T ( u ) 3 . Rarh n ~ r m n t a . t i n nin t h e sum contributes a term of the form

.

,1

is 1 when the sum is expanded in terms

of the standard basis for the Temperley-Lieb algebra. The proof we present follows [18].

88
?.

THECLASSICALD QUANTUM AN 6j-syMBOtS

UANTUM

~1(2)

89

A - ~ ~ to ~ ) coefficient of ( the the computation
!.',

,1

by the bracket identity.

we need

3.5.5 Remark. The formula given above for the projector in-

icates precisely how ing projection =
A*~T(o).
a€Cn

4 is analogous to the classical symmetrize :

4.Moreover, the classica~projection satisfies the ,

i<

same recursion relation with quantum integers replaced by inte-

gers. such a replacement is an evaluation of the quantum Projector at A = 1, and for that value of A, positive braiding is indistinguishable from negative braiding. When the Jones-Wentzl projector +,A defined ( e - g . whell A is is transcendental), it is a nonzero central idempotent such that ~ + 1) 4

To prove this, we use long multiplication to multiply out
I
(A4(,-')

A ~ ( ~[ n ~= - ~! )

-

+ A4tn-') + . . .

+ A ~ ( ~ + .~. .)+ 1) . . .( -

4 .IA(n) = (0). So the Temperley-Lieb algebra has a direct sum 2
decomposit~onas scalar multiples of the Jones-Went21 projector and the ideal I A ( ~ ) : TL,(-[2])
=

The terms in each of the factors are arranged along the horizontals in the triangular array that is depicted above. A term in this expansion corresponds to an ascending path. Moreover, the coefficient of when like terms are combined, is the number of paths that have k: points to their right, and tllis is the number of
r ) ( : ~ , t ~ l ~ ~ t a t 1 1 a t 111itti111i111,~ intis 1 IISO

c+!(33 Ia(n).
d 1,; ,

lve have the direct sum decomposition
1 1 2) (\,i,.:.,,
- +;A(\,;/~),<JIL -

1; ;i(~,ji,c(tllt

I , . ~ , ~ , 5 , ~ ~ ~ s j ~ j ~ l ,.~.i,,~,> ,5~

coefficient of

I,

in our expressioll for

4 is 1. :

Now for any given k, the minimal product representations of the Permutations can be chosen in such a way that the set of all permutations is partitioned into a set of words W that do end wit11 a k = (k, f I , and the set Wok. (For example, ) the triangular scheme above does this for k = 1.) Clearly, these two sets have the same number of elements, and the number of transpositions that it takes to write an element wak is one more than the number of transpositions that.is takes to write w E W. cause 6khk = --A3hk, and the contribution of word w E
for

for representations of B(n) and Up(sl(2)) where
112 L; = I A ( ~ ) ( V A)8"

is the kernel of the projector
3.5.6

4 This leads us to the following: : .
;":

/

i11

Definition. A Uq(s1(2))-equivariant map
$j

i

= df

V :

-

112 (vA )8 2 j

I

is defined in terms of 'the projector as follows:

The computation that 0 = ( ~ , , z n ( ~ - 3 ) T t u ) d )hk f o ~ o W S beSO

4,(e.
3

3,m

j = A(j+m)(j-m)+A.(x~~+my ~ j - ' n ) . 8
23

w is

This formula only makes sense when [2j]! fined, the map
$j

#

0, and thus the

canceled by the contribution of wak. A similar argument that 0 = hi, ( Z , E ~ ~ ( A - ~ ) T (.~ ) ~ ) completes the proof. This

value of the quantum parameter is important. When it is deis a lift of the map w j (which was defined in

90

THE CLASSICAL D QUANTUM AN SYMBOLS

) QUANTUM

d(2)

Theorem 3.4.1); in other words, pjq5j = wj where pj : (v1I2)@2j _,

( V ~ / ~ ) @ ~is the projection. ~/LT
The proposition that follows indicates that the image of $ j is the quantum analogue of the symmetric polynomials in x and 51.
3.5.7 Proposition.

I

3.5.9 h m m a .

For a

+ b = n,
-

where

where

t , ( S ) is the minimal nunzber of adjacent tran,spositions that it

for R =

~ ( I L ' T . ~112

~ ~ ~ o ." ~ ! m - ! / ? r . ~ i ! / , s r / i c

fn

-L

1 . . . ..!! -!.-1):

(, j 1 , ; . .

!
1
.
#,

P r o o f of Proposition 3.5.7. .4ssrilning Lemma. 3.5.9 nrc.indrlct.

on b. In case 6 = O the result Sollows by usi~lg inductioll on ,n a.liti
X

$

the recursion relation for the projector 4 . : For general values of b we use the maps

4j : V i

4

I/2 823 (VA )

defined above, where j = n/2. On the one hand,
q5j

k
i $

(Fej,m> = F43 (e3,m) = A ( J + " ) ( ~ - ~ )n~ ( + A ( ~ @ J + ~ 8 y@3-m

5
where
dr

>1
On

+ is evaluated by means of the inductive hypothesis. i

the other hand,

$J(Fe,,m) = h ( Z j m ] e , , m - ~ ) q = [ j + m]~(j+m-l)(~-m+l)+~(x@j+m-ly@~-m+l n 1.
The proof will follow by comparison of the two sides of the equation once we establish the following:

+

Proof. We have for example,

q P p l yLemma 3.6.1 to complete the proof.

1 3.6.3 Definition. n-1 n-1

Let U=UA= U* : C 4 n @ vii2. Havn ing defined U = U A : C 4 (v:/~)@~("-'), define U = U A to be the 112 f i . 3 v,1I2a ...@VA 5 ~
n-1

1

1

1 ' composition
3.6.2 Lemma.

C
R C {1, ...,n}

q-2tn(R)

= 9 -k(n-k) The map
n n n=nA is defined dually, and is also in analogy

with

the classical case.
3.6.4 Lemma.

&dl)=

( ~ A ) " - ~ I S I.~ .~ @

Z.s

32

. . .@ f

where tn(R) is the minimal number of transpositions that it takes to move the subset R to the subset {n

+ 1 - k, . . .,n).

where

Proof. The proof will follow by induction. We indicate the proof
of the first formula; the second follows similarly. and

Also we have,
((x18..-@xn)@(~n@...@Z1))

.

.

={O

if~~=~~forsomeIc=l,...,n (iA)n-2S if {xk, Zk) = {x, y) for k = 1,.. . , n

where xk, Z k E {x, y} for all k = 1,2,. . . , n and s = #{k : xk =

Y. )

96

THE CLASSICAL D QUANTUM AN SYMBOLS

QUANTUM

sl(2)

97

Proof. As in the classical case, the proof follows by induction

I
3.6.6 Theorem. (1) Let A E C. Let (a, b,j) denote an admisk

(See Lemmas 2.7.3 and 2.5.3). In case n = 2 the non-zero values are as follows:

sible triple such that if A is a primitive 4rth root of unity, then

1 max {2a, 2b, 2j)
(v:'~)"~'

<

r. Then the Uq(s1(2)) invariant map

y

:":

-+ ( v ; ' ~ ) @ ~ ~~~ @ v~ )' '( :

1 8

is defined and

I

fi(y@y@x@x)=(i~)-~.
j

3'6'5 Definition.

c

i(b-~)-(a-~)A(b-u)(btutl)-(a-u)(a+~+l)

I
Let j E {0,1/2,1,3/2,. . .). Define a map p j : ( ~ j ' ~ ) @ vj via +~ j
/cj(X,'

u+v=j

(2) Let A E C . Choose a,b E {0,1/2,1,3/2 ,...I so that if A is a primitive l r t h mot of unity, then 2a t 2b < r. There is a direct
sum decomposition

@ ~2 @ . . . @

2 : ~= X I ~ )

. X2 . . ..

'

~

2

j

?! ,I
where the lnultiplication on the right occurs in the ring of polynomials in non-commuting varia.hlps -r7 3 ?,- F . ~ ; JJ Pj. ;" ,"q\OO\'=')) ~ ~ ~ ~ ------7 ,, ~ + rr / ,J/Q\\ map and its kernel is L?.
Q
-"-"LL"L

@

lfj = @ /LC, CJ
I

/lf,

where the sum is taken over all j such that ( b , j ) is 7 ' Furthermore, if (a, b, j ) is admissible, then any Uq(s1(2)) invariant map V -+ V ; ;
@

Next we mimic the classical case to define, for (a, b, j ) an admissible triple of half-integers, a Uq(s1(2)) invariant map

V is a scalar multiple of Pa @ Pb ;

Proof. Recall that ej,t denotes the weight vector in Vi of weight

,~ ) t, and that ~ ~ ( e i= A ()J - " ) ( ~ + ~xjtm
ab

b-'.

The restrictions on

via the formula

and j insure that

J

where Im is the identity map on m tensor factors of Vi/2 (Cf. Section 2.5.4).

-

Ima 3.6.2, and Proposition 3.5.7.

v . '

'' i I

The formula for pa @
is defined.

ab (4,

(ejj))) follows by computa,tion using Lemma-3.6.4,
/

i

1s

defined and non-zero for all 3 such that (a, b, j ) is admissible and that the representations Vi are irreducible since j 5 a -k b < r / 2 *

i Ir
I
i

The argument that V j @

VA

splits as a direct sum of the images

3.6.10 Lemma. The quantum Clebsch-Gordan coefficients sat-

i ;

of the V j follows the same lines as in the proof of Theorem 2.5.5.

isfy the following recursion relation

3.6.7 Definition.

There is a Uq(s1(2)) invariant map hlb: -+ ( ~ ; ' ~ ) " ~ j defined for admissible triples , Hence,

(v;'~)"~~@ (a, b, j ) as follows:

I

l
II

(4, 8 & : V; 8 V + V j is also , ) ; Uq(s1(2)) invariant, and it corresponds t o the projection of V$@\<
o
ab

The composition pJ o

,/,
3

i

onto the direct summand that is isomorphic to Vj.
3.6.8 Lemma. For u t v = j, (a, b, j ) admissible, and max 120,

(- l)z~(z-w)(i+t+l) [ a + u + z]![b v w]! [z]![w]![a - u - z]![b - v - w]! z,w: z+w=~-t The sum is understood to be over all integers z, w such that z+w =
j - t and all the factorials are of non-negative integers. Further-

+ +

I

2b, 2 j )

< r if A is a primitive 4rth root of unity,
/
1
\

7ue

have

; more, u, v, and t are weights of V$, v, u+v=t.

and V;,

respectively, and

by applying F t o the equation that defines the Clebsch-Gor dan
i

coefficient. The closed form is determined by solving the recursion using the value
a , b . j - i(h-u)-(a-u)A(b-~)(b+v+l)-(a-u)(a+u+l) Cu,v,~ -

Proof. The computation is analogous t o the proof of Lemma 2.7.4,

but relies on Lemma 3.6.2 and Proposition 3.5.7 to take care of the powers of A.
3.6.9 The quantum Clebsch-Gordan coefficients.

[a + - jl! [a - u]![b - v]! '

We have maps pa 8 pa

0.

Y

ab
0

3.7 Quantum network evaluation.

Recall that in the clas-

4i

: V -+ V j @ V when the ; ;

triple (a, b, j') is admissible. Define the quantum Clebsch-Gordan coeficient CZ;;$ to be the coefficient in the sum

i

sical.case the key computations were the ev,aluations of the closed . "theta-nets" and the closed "tetrahedral" networks. In [18], these network evaluations are given in case A = ei"l(2~) and their diagrammatic computations can be applied to the case of generic values of A as well. We use the machinery developed above t o give alternative computations.

100

THE CLASSICAL D QUANTUM j AN 6

-

s

~

~

~

~

~

~

3.7.1 Theorem. Let A E C . Choose a, b, j, k E {0,1/2,1,3/2,.

.

such that
1. (a, b, k) and (a, b, j) are admissible triples;

2. I A is a primitive 4rth root of unity, then assume that f

i
t

a, b, j, k < r/2.

(
L

a+b-j a-u

1
J

$a(ea,u) 8 $b(eb,v))

)
+ vl!

where 6 is a Kronecker S function, AJ = (-1)2~[2j + 11, and ; @(a,b, k) =
1

= e3,3(-l)aSb-3
-

([a

+ b - j]!)"
[2a]![2b] !

I .

\ I . .

q(b-'JJ(btvtlJ

. . \ _ 4 ul![b la -

U+V=J ( a - ~ ) ( a + ~ + l ) u]![b - zlj! q

r -

k,,

\-1)

,

,T, [(I

+ 1)

-

1:lt[rt

-

b t X I t [ - ( 1 t b -1X ] ~ [ O b + 1 4- 1 ' 1 + , [2a]![2b]![2l;]!

e3,3

(-l)"+b-"

[a + b - j]![a

+j

- b]![b j - a]![b [2a]![2b]!\21 I]!

+

+

+ j t a t I]!

'l'llo last ecluahty 1s a

( ~ U ~ I I ~ U I O I ~ I ~ I 1C I~ ~ e ~U t\ i f111~11 C ~ ~ ~ d ~ l I

3.7.2 Remark.

analogous to the identity used in the proof of Theorem 2.7.6. Thus In case k = j = 0 and a = b, the formula (-l)""[Za

reauces to

A, =

+ 11 and this 1s the value of a closed
L

t o complete the proof we need use the following:
3.7.3 Lemma.

8%

value O(a, b, k ) is the value of a closed network as in the ~ a r"a ~ r a ~ h3 -

C %
:

u+v=j

-q i L q(a-u)(a+u+l) [a - u]![b - v]!

Proof. By the assumption on A, the representations V; and V:

[a + j - b]![b$ j - a]!
Proof. The proof is a yantization of the argument that we gave

are irreducible. Ther and when k = j it is a multiple of the identity. In Theorem 3.6.6 weight vector, and in these results and manipulate the expression for the exponent of A while recalling that A2 = q and obtain the following:

during the proof of Theorem 2.7.6. We rely on the quantum combinatorial identities stated in Lemma 3.6.2. Furthermore, we recall
from the proof of Proposition 3.5.4 that

c

qk2T(u)

= A*"("-1)

[I.!

aECn

102

THE CLASSICAL D Q U A N T U 6j-SYMB AN M
an every element of f ((1,.

103

where T ( a ) is the minimal number of transpositions of the for (k, k 1) that it takes to write a.

+

..,a + j - b))

merit of f({a + j - b + 2,. .., 2 j 1)). We evaluate
two ways to obtain the desired identity-

+

and less than every

LEB q

2T(u)

Consider a sequence 0 = a0

< al < . . . < a, = n, and I

f

: {I,.

. . ,n)
T

+

(1, . . .,n) denote a bijection for which the

,

on the one hand, when S C { I , . . - a + + j +
7

is a subset

strictions f Thus when

+ ,. . - ,aP+l) are increasing for 0 5 p 5 = 2, the permutation f acts like a shuffleto a deck
+ 1,. . . ,ap+ll,
B~ ={a E B : a ( { 2 j + 2 , . . . , a + b + j + l ) ) = S } .

cards- An arbitrary permutation up of the set {a,

is extended to a permutation of 11,. . .,n) by the identity. ~h~~

compute the number of adjacent transpositions that it takes to write f 0 (goal . to obtain T(f (0001 . . . a ,-I)) = T(f)

+xT(ap).
p=o

T-1

- +b + For such an S, let fs be the unique permutation of + 1) such that fs E Bs and the restrictions f \{I,. . . ,23 1) are both increasing. andfI{2j+2,...,a+b+j+l} The set B can be written as the disjoint union of the sets Bs

+

as

s ranges over

subsets of size a

+ 6 - j- Furthermore,
x{l,...,a+j-b),

(The preceding formula follows rather easily from the depiction of permutations as strings crossinq in the planc.)
111 a si111ilar f a s l ~ i o ~JF1Ii if' ~

B~ = {fso (alanaa)

: 01 E

O<p<r-1,then

is ~ p + 1 , . . . , ~ p + l } increasing for

From this formulation we have,

If f : {I,...,n)
creasing, then

+

{I,.. .,n) is a bijection such that the

restrictions of f to (1,. . . ,m) and to {m

+ 1.. .,n) are both inwhere the exponent is given by

N = (a+j-b)( .+j-&l>r(b+j-a)(b+j-aa1M
because t,( f ({m standard position. Consider the set, B , of bijections f : {1,2,. . ., a

a b-3) (a b-j-1
-

+

+

+ 1,. . .,n)))

is the minimal number of trans-

positions that is needed to put f({m

+ 1 , . . .,n))

Next we sum over all the subsets S to obtain

back into its

+ b +j + 1)

-+

{1,2, ..., a + b + j + l ) s u c h t h a t t h e v a l u e f ( a + j - b + l ) i s g r e a t e r

1

i

104

THE CLASSICALD AN

QUANTUM

SYMBOL^

E
On the other hand, we compute

the number required to move S into the final position, i.e. t(S). Recapitulating, T(ge,s) = ( - a - j C

cEB( examining the ~ ~ ~ by " 1
+ j - b < C < 2a, let

'
F

i

images of the various a. Specifically, for a

+ b)(a + b + j + 1- C) - t(S).

We recall from Lemma 3.6.2 that Then B is the disjoint union of the Bes.
F o r a + j - b I C I 2 a , a n d f o r S ~ - { 2 j t 2 , . .. , a + b + j + l )

of size I 1 = C - (a t j - b), let S

Let ge,s be the unique bijection on {1,2, . . .,a that ge,s E Be,s and the restrictions of and {C
( l ~ , i o ~ i l~l il i

+ b + j + 1) such

Putting everything together, we get

+ 2 . . . . . n + b + j + 1) are both increasing. Then Br is the
$ 1

giito each of (1,. . . , e l

*IS1 IIP /I,

,. < \ I ~ ( Y

As before, we examine the number of adjacent transpositions required to write various permutations, and obtain
q 2 ~ ( u ) (q2~(ge,s)) =

= AP

C q2e(.i+l)
e

[[]![a

+ b + j - C]!

(=

q2~iul))

.(

q2~(u2))

where the exponent E is given as

oEBe,s

= ~ ~ q ~ ~ ( g e , s ) [+ b ! [ j - C]! C ] +a where theesponent M = C ( C - l ) + ( a + b + j - C ) ( a + b + j - C - I ) . observe that T ( g e , ~ ) the number of transpositions of the is form (k, k move {a {I,. . . , a

and the related exponent P is the following
-.

P = (a+b+j)(a+b+ j-1)-4( a

+ j - b)(b + j_+1).

+ 1) that are needed

to move {a

+ j - b + 1 , . . .,C}
for

Set e = a+u and set j = u+v, and compare the two expressions

into the set S. This is the number of transpositions required to

+ j - b + 1 , . ..,C)
+b +j +

CuGB We get q2T(a).
q2("+")(j+l)I" + ~ l ! [ b [a - u]![b u+v=j
J -

into the final position in the interval I}, which is IS[(a b j 1 - e), minus

+ + +

"'"

106

THE CLASSICALD AN

QUANTUM

SYMBOL^

Lemma 3.7.3 follows by manipulating the exponents. This corn-. pletes the proof of Theorem 3.7.1.

for various values of j. Second, consider the composition

3 7 4 Remark. ..
@(a,b, j )

#

In case A is a primitive 4rth root of unity, 0 precisely when a b j 5 ( r - 2). We will give the

+ +

representation theoretic meaning in Section 4. In this section we define the quantum 6j-symbols in case A is not a root of unity, and verify that the Elliott-Biedenharn identity and the orthogonality identity hold here. In Lemma 3.10.9 we establish the diagrammatic rules that are necessary to prove the analogues of the identities give11 in Theore111 2.7.14. esar-tlinc the svmWe
3.8 The q u a n t u m 6j-symbols - generic case.

for various values of n. The values of j and n are restricted so that (b,c, j), (a, j, k), (a, b, n), and (n, c, k ) all form admissible triples. Alternatively, if one of these triples is not admissible, then we declare the corre\

I

I

108

THE CLASSICALD AN

QUANTUM

SYMBOL^
and

!
!

I

3 8 3 Definition. ..

coefficient

{

}

Define the quantum 6j-symbol to be the in the following equation.
4

Observe that these calculations are analogous to the classical case. Moreover, the recursive method given in Section 2.8 that is used to compute the values of the 6j-symbols in general also applies to the quantum case.
3 8 4 Identities a m o n g diagrams. The key to simplifying the ..

computations of the classical case was t o derive identities among the diagrams that represented various maps. To this end we observe that Theorem 3.7.1 gives the evaluation of the closed theta network for certain values of (a, b, j). The identities that are expressed in Lemma 2.6.4 and their proofs go over to the quantum case as theg are stated. The essence

( a , j , k), (a, b, n), (n, C,k) is not admissible. For example, consider the case when a = b = c = One can compute directly from the definitions that
= 112.

The matrix on the left represents the maps U and n, and this element squared represents the composition of the operators fl €4

3 8 5 T h e o r e m (Orthogonality). ..

Suppose that qr
.

#

1. The

quantum 6j-symbols satisfy the following relation:

Proof. Identical to the proof of Theorem 2.6.6.
3 8 6 T h e o r e m (Elliott-Biedenharn Identity). ..

The follow-

ing relation holds among the 6j-symbols when qT # 1.

THE CLASSICAL QUANTUM AND 6j-SYMBOLS

UANTUM

~ l (

Two tangle diagrams are regularly isotopic if one can be obtained from the other by a sequence of the type I1 and type I11 Reidemeister moves. More precisely, when a height function is chosen on the plane of projection, then the 3-dimensional analogues of the identities expressed in Lemma 2.3.2 together with the diagrammatic representations of the identities s,syl
= 1,

-

sjsj+l sj = sj+lsjsj+l, and their obvious variants obtained by
changing appropriate crossings are the moves that generate regular isotopy of tangle diagrams. These last two moves are depicted below. Thus a regular isotopy is a sequence of these diagrammatic

Figure 6: A tangle diagram
3.10.1 Definition. A tangle diagram is a diagram of a collection

moves, and each such move corresponds to an identity between

f

maps represented by the respective diagrams.

of knotted and linked proper arcs and circles in a rectangular box in which the arcs in the diagram terminate on the top and bottom rectangular faces of the box. The box is projected to the plane of the paper, and the diagralll is a.ssiuucd to b(!
111

I

l

l

~ c ~ l c r posibion al

meaning that a t most two arcs cross at a given point, and such a crossing is transverse. The diagram depicts over and under crossing information in the standard way: a t a crossing point, the arc that is farthest from the plane of projection is broken. An example of a tangle d.iagram is given in Figure 6.

A tangle diagram for which n arcs come out of the bottom of the diagram and m arcs come out of the top of the diagram represents a map
T:(V, 112 8 n (Vl/2)8m 1 , We have established the following:
3.10.2 Lemma. (cf. Lemma 2.7.2). Regularly isotopic tangle diagrams with n strings at the bottom and rn strings at the top

via the association of maps n and U to generic maximal and minimal points and via the association of the bracket identity to each crossing. In the sequel, we will identify the diagram and the represented map.

represent the same map
T : (v:/~)@~-+ (v:/~)@~.

114

THE CLASSICAL QUANTUM AND 6j-SYMBOLS
Let a, b E (0,1/2,1,. . .). Define maps

115 Suppose that 2b strings come out of the top left edge of the tangle S , suppose that 2a strings come out of the top right of S, and 2 j strings come out of the bottom of S. The top-right, topleft, and bottom of the tangle are called the a , ,f3 and y regions of the boundary, respectively. Thus S represents a map ( V ; I ~ ) @ -+ ~~

3.10.3 Definition.

I;:(+)

"nd I;:(-)

that are quantizations of the map that switches

factors defined in Lemma 2.7.5. Thus,

The (unquantized) switching map

X : t

can be written as a prod-

(v;/~)@~'@ (v:'~)@~~.The strings that come out of the coupon are represented by the three ribbons in the picture below. We will compare the composition Kt:(+) tion (Tz,(-) @ T2b(-))
0

uct of adjacent transpositions since it represents a permutation; choose a minimal such product. Each transposition uk ='(k, k t 1) in the product is lifted to a braid generator ci,, = &$(+), and it is lifted to
&-I

o

S to the composi-

sk

to define Hence, in

2 0 Tzj(+) where each of these tangles

= s i l to define Kt:(-).

is associated canonically with a map among tensor powers of the fundamental representation.
3.10.5 Observation. The diagrams depicted below are isotopic

the diagrammatic representation of Kt:(+)

the 2a strings on the

left cross over the 2b strings on the right as one reads from top to bottom; in Xt:(-) the strings on the top right cross over those on
the top left.

but not necessarily regularly isotopic diagrams. The isotopies
.1/2..>,

fin:

])cfilLe i~kapsI ; , , ( & ) : J -" 1 tl~i1.1,i i ~ . ~ ' I.O~II'C~sented by half-twists in the cable of n-strings. Specifically, the half-twists can be spelled out in terms of braid generators as the products: Tn(-) = (S;~)(S;~S;I) while Tn(+) = (sl

11.2 . . , L , ,

//I(

((I</, (I-f !h(
. q

O I I / ~ O ; I I / ~~ < / i ~ t ? , % ,

- .. (S,~~S,:~-

.ST')

Z

. . .sn-2~,-1)(sls2 - - .sn-3sn-2) . . - ( s I s ~ ) ~ I .

To obtain the desired maps, It:(&) and Tn(f) apply the bracket identity t o each braid generator in the product. 3.10.4 Notation.

Suppose that S is a planar tangle with no
Demonstration.

closed loops. Thus S consists of properly embedded arcs in the rectangular region containing S. We depict S as a rectangular coupon as in the illustration on the left below. The tangle is a 180' rotation of S through a vertical axis.

Cut out a piece of paper in the shape of a

thick Y with fairly long edges. Arrange the edges with a crossing as in the left hand diagram and tape the ends to a table. Write

THE CLASSICAL D Q U A N T U M AN 63-SYMBOLS
the letter S on the neighborhood of the vertex of the Y. Then rotate the vertex about the vertical axis. This gives the isotopy. The maps that are represented by the diagram on the left and the diagram on the right differ by a factor of ( - A ~ ) ' ~ , for such a factor measures the difference between isotopy and regular isotopy. We next quantify that difference. In the planar tangle-S (with ,8 region on the upper left, a region on the upper right, and y region on the bottom edge) suppose that
I

QUANTUM 2 ) 4

f

and

6 OX.(*)

= (-aA3).

o(Tn(+) 8 Tn(+))

where X,(f) denotes n strings crossing from top left to lower right over/under n strings running from lower left to upper right, respectively.
Proof. Bracket aficionados will recognize that

X(f)u = (-A)*~u and nX(k) = ( - ~ ) * ~ n ,

x strings start and end in the p region, r strings start and end in the a region, p strings start and end in the y region, w strings
run between the ,8 and the cr region, u strings run between the

p

and y regions, and t strings run between the cr and the y region. Otherwise the diagram for S can be quite arbitrary, and there is ~ i o I ~ ~ ~ ion I ~ 1 ) o 1 111c 1 1 c ~ \ 1I I P , , II II ~ 1t I
01 i ~ l f ~ l . l i ~ ( i11g

"with signs read respectively. (Non-ahcionados are encouraged to
111 follows I,y ind~~ct.io~t.

-

of

~ ~ P S tF i l l q \ ~ r

beyond the condition that no strings cross. The tangle Z denotes the map represented by a rotation of S by 180' about the vertical axis.
i

;d

Tlle LI.ILI\
A ,

L,

01

Lne alagram. u
,a

1s

to l e g u I a l I y

15o~opc

The diagrams representing Xn(f) ?J and

fi In(&) will be called

3.10.6 L e m m a . In the notation above

For the case in which x = r = w = a = b = 0 and p = n we have the first equation below. The second equation below corresponds t o the case when all strings start and end in either

Proof. We establish some special cases then we will proceed to

the a or the /3 region; i.e. j = 0 and either a = 0 or b = 0. 3.10.8 S u b l e r n m a ( c a s e 2).

1
I
t

i
I

the general result. Corresponding to the case x = r = p = u = t = 0 and w = 2a = 2b = n in the notation above, we have the first equation in the following: 3.10.7 S u b l e m m a (Case 1).

?I= (

- A ~ ? ~ ~ I) oT2,(f)

6=( - A ~ ) * ~ T ~ , ( + )I? o
where the signs are read respectively.

118

THE CLASSICAL D QUANTUM AN 6j-SYMBOLS

Proof. We induct on n. The case n = 1 is the same as Case
1 above when n = 1. For larger values of n, one small kink can

be canceled from the diagram representing the map on the right (of either equation) at the expense of multiplying that factor by

(-A~)*~. Because T represents a complete half-twist, the arc from
which the kink has been canceled can be isotoped away from the rest of the diagram. We turn now to the proof of the general result. Let S' denote the tangle obtained by twisting the w curlicues out of )$:(+) Specifically, in each string that runs between the
CY

o S.

and

P regions,

the small loop is removed. These loops are removed successively. Thus we replace Xw(+) o Uw by (Tw(-) @ Tw(-)) o U Then , .

as in Case 1. Similarly, let Z' denote the result of removing the curlicues from the
T,X,

and p strings that run back and forth to the a,

p,

and y regions, respectively. The nesting among arcs that run

back and forth to a given region is not relevant, as can be seen by examining the figure below. In this way we obtain, Furthermore, Z' and-S' are regularly isotopic, and thus represent the same map from ( ~ i / ~ ) ' @to j( ~ i / ~ @ (v1I2 )8% ~ ) @ ~ ~ . A

(T2,(-) @ T2b(-)) o Z o T2j(+) = ( - A ~ ) ~ - ~ - ' Z '

)$:(+)

0

S=(-A~)~s'
0

= ( - A 3 ) ~ ~ ' ( - A ~ ) ~ + " + T(T2a(-) 8 T2b(-)) = -P by Case 2. This completes the proof.

0

T2j(+)-

1

120

T H E L A S S I C AAL D Q U A N T U M ~ - S Y M B O ~ ~ C N ~

3.10.9 Lemma. The following identities hold among the Uq(s1(2))
invariant maps.

Proof. \,\'(,

115t

L!tl~iil~a. 3.10.6

I\!(. t.o provc it.cn.1 I as follo\~~s.

restrict to the case of Xi:(+), negative crossing case is similar. the The map

Y

ba

can be written as a linear combination of maps

f o h v s . Expand each of the projectors

%a, 4tb, +fj in terms and

of the bracket identity. Replace any closed loop that might appear in the expansion by the factor -A2 - A-2. Thus we have

where f s ( A ) is a polynomial in Af tangles S with 2b

and the sum is over all planar

+ 2a outputs on the top and 23' outputs on the

bottom. Next we consider the exponent on the factor -A3 on the right hand side of 3.10.6. We have x+r+w-p=a+b-j

122

T H E L A S S I C AALN D Q U A N T U M j - S Y M B O L S C 6
Item 2 follows from item 1 by techniques of regular isotopy. Items 3 and 4 follow trivially from the invariance of the map represented by regularly isotopic diagrams.
3.11 Symmetries.

regardless of the planar tangle S. This gives

Here the twisting rules will be used to

adjust the 6j-symbol to one that has full tetrahedral symmetry. because the projector

+,A invariant under is

180' rotation for n =

3.1 1.1 Lemma. The quantum 6j-symbols possess the followiny

2b, 2b, 2j.
The second step of the proof is now relatively easy. Observe that

symmetry

Tn(-) 0

+;

= A"("-')/~+A

n-

This follows by-applying the bracket relation to each of the negative crossings in the half-twist. In the bracket relation either stra.ig11tstrings or lnasilnum and minimum points result. The opannihilated by the Jones-Wentzl projectors. Consequently, all of these terms with a coefficient of A-' vanish in the bracket expansion. Similarly we compute

Proof. Identical to Lemma 2.7.8. U
3.11.2 Lemma.
\

where the functions @(-,
Combining results we have

-, -) and A are defined as i n Theo-

3.11.3 Lemma. The symbol,
a+b-j~3(a+b-j)~a(Pa-I)+b(2b-l)-j(2j-l)

-

= (-l)a+b-i~2(a(a+l)+b(b+l)-j(j+l))

This completes the proof of item 1.

THE CLASSICALD QUANTUM - S Y M B O L S AN 6j

where a choice of each square root is made once and for all, is invariant under all permutations of its columns and under the exchange of any pair of elements i n the top row with the corresponding pair i n the bottom row. Equivalently, the symbol is invariant under the permutations of the set
3.11.5 Remark.

[: t ]

where Ak = ( - 1 ) ~ ~ [ 2 k 1 . 1
Proof. This is a direct computation. EI

+

9

These identities are slightly different than

{ { a , b, f

1,{ a , c , d l , {b,c, e l , {d, e, f }1.

Proof. The proof of Lemma 2.7.12 needs t o be modified as fol-

they appear in the classical case. In that case, our choices of the b c j symbol were motivated by the desire to make conr

-

L

lows. When the closed tetrahedral network depicted in that proof is rotated through space to reverse its orientation, four twists at the vertices are added. Each of these twists contributes a factor of An where n , by Lemma 3.10.9, depends on the labels at the vertices. A careful count indicates that these powers of A cancel.
n U

tact with the existing literature, specifically [2]. In the quantum case, these identities coincide exactly with those in [32].
3.12 Theorem.

-

In [24], Masbaum and Vogel use the recursion relation for the Jones-Wentzl projectors t o give a formula for the quantum 6jsvmbols. See also Section 2.8.
3.11.4 Theorem.

A.,

The orthogonality relation and the

Elliott-Biedenharn relation hold for the normalized quantum 6j74 coefficients i n the following form.

b

Orthogonality,

Elliott-Biedenharn:

THE CLASSICALD QUANTUM AN 6j-SYMBOLS

=

l)k+c+n+m E(n

A2(k(li+l)+c(c+l)+n(n+l)+m(m+

{ ;} , { } , { ;1 , .
3.
12a12b

=

z3@ ( a ,b, j )
b, k)
- 'y(-1)a+b--3
3

.

4
1)).

The Quantum Trace and Color Representat ions

In this section we will define the quantum trace and use it in the case where A is a root of unity to distinguish representations that are not used in the topological applications that we have in mind. The representation theory of Uq(s1(2)) in the root of unity case is worked out in [20] under the assumption that E and F are nilpotent and I(; is of finite order where the orders are determined by the value of q. We will not need to make that assumption as it

4.
(-1)
a+b-kA2(k(k++l)-a(a+l)-b(b+l))~,

holds on a l the representations that we consider, nor do we need l to classify all of the representations of Uq(s1(2)).

,12(a(b+l)+b(b+l)-1(~+1))@((~,
' 2

b13)

{UZ }

4.1 The quantum trace.

map f : W

-

We will define the trace of a linear
15

147

in case Mr

a lT,(s1(2)) ~epresentation.This
a.1-

notion of trace is quite versa.tile. In particular, such a, tra.ce

Proof. The proof follows along the same lines as in the proof
of Theorem 2.7.14, but crossings are lifted to over and under crossings in a consistent manner so that the diagrams can be isotoped in 3-dimensional space. The remaining details are left to the reader.

lowed Jones [12] to define an invariant of knots via braid group representations. Lickerish's [23] combinatorial definition of the Reshetikhin-Turaev invariants [29] is given via the trace. Lickorish's techniques are employed in Section 4.2 in order to prove that the matrix representation of the Temperley-Lieb algebra is faithful for generic values of q. Finally, we will use the trace to characterize certain represen-

1I
I

0

3.12.1 Remark. Similar identities hold for the normalized 6j-

I

symbols as well.

g

tations in the case when A = eZ"/(2T) any other primitive 4rth (or root of unity). The tensor power (vif2)@Indecomposes as a direct sum of a trace 0 summand and a sum of representations Vi where
j is taken from the finite set (0,112,

. . . ,( r - 2)/2). By excluding

the representations that have trace 0, we will be able to define the Turaev-Viro [32] invariant as a state summation.

128

T H ECLASSICAL D QUANTUM AN 6j-SYMBOLS

T H EQUANTUM TRACE N D COLOR A REPRESENTATIONS129

4.1.1 Definition. Let W be a representation space for Uq(s1(2)).

Then the quantum trace of a linear map f : W nary trace of the map K 2 o f . In particular, if

+

W is the ordi-

L : (vy2)@n 112 )@n -+ (VA
denotes a linear map, then

tr,(L) = t r ( l i 2 o L ) .
Consider the bases { f s : S C {1,2,. . .,n ) ) and

". .- '

{fs

:

S

c

{1,2,. . . , n ) ) for ( v : / ~where ~ = xf 8 ) @ f, z,S 8 . . .8 ~f with
XjS

- - . xz 8

and

f, =
~ ~

Figure 7: The braid closure of a tangle
Proof. The second equality follows because Ir' fs = A ~ ( ' / ~ - " / ~= fs )

=

x if j @ S
y

- fS~ where Tl = # { k : x z = x ) and s = # { k : x: = Y 1 ~ ) .

For

if j E S
if ~

the first formula, wc compute using Lemma 3.6.4
S

and

-

( z

F

i4 o ( L @ In)o ilr ( 1 )

=

ij o(L 9 1,)

1

and the sum is taken over all subsets R c ( 1 , . . ., n).
= (-1)n-

1

4.1.2 Lemma.

C
S C { ...,n } ~,

~ 2 ( n - 2 1 s l ) ~ ~ ~

This completes the proof.

I

S

where the sum is taken over the set of all subsets S C {I,. - .

."1.

<

4.1.3 Remark. The point of the Lemma is, for maps L defined

on ( v ; / ~ ) @ ~ terms of Temperley-Lieb elements, the trace of the in

130

T H ECLASSICAL QUANTUM AND 63-SYMBOLS

T H EQUANTUM TRACE AND COLOR REPRESENTATIONS 131
the association of the bracket identity to each crossing. Regularly isotopic diagrams where the end points are fixed during the isotopy determine the same maps. Let

operator can be computed by the bracket expansion of the braid closure of the operator; the braid closure of a tangle diagram is depicted in Figure 7. In particular, we apply the Lemma to the Jones-Wentzl projectors to obtain the following:
4.1.4 Proposition. Suppose that A4T # 1, or if A is a primitive

I denote the set linear combinations of such tangle dia,

grams. The bilinear form

4rth root of unity, then suppose that n

< r - 1.
is defined as the h e a r extension of the map defined on tangles S and M by:

In particular, if A is a primitive 4rth root of unity and i f n = r - 1, then trq(C-")= 0.
Proof. The proof follows from comments made in Remark 3.7.2.

(S,M) = o(S @ M ) o ir ;

where W is a rotation of the tangle M through 180" about an axis perpendicular to the plane of the dia.gra.m. The second cqua.lity follows because the diagrams representing these tangles a.re regularly isotopic. The image of (., -) is C by evaluating the diagrams via the bracket identity; the parameter A is assumed to be a nonzero complex number. The Temperley-Lieb algebra, TL,, has dimension (:)/(n+ 1).

4.1.5 Definition.

and Now suppose A = e"il(27'), as usual,, let

trace 0 if trq(f ) = t r ( K 2o f ) = 0 for every Uq(s1(2)) invariant map

4.2 A bilinear f o r m o n tangle diagrams.

Recall that a

A vector space basis is a certain set of monomials in the generators
(1, hl, - -, h,-I}. Because of the relations among the products of these generators, any monomial can be put into a normal form.

tangle diagram is a diagram of arcs embedded in a thin rectangular prism obtained by projecting in the short direction onto a rectangular face and depicting crossing information by breaking the arc at an intersection point that is farthest away from the plane of projection. Now suppose that such a diagram h s the same number of incoming and outgoing strands, say n. Such a tangle diagram (in general position wi.th respect to a height function) represents a map

We will not list the normal form monomials here, but we will assume that these form a basis. We can mod out the free module generated by regular isotopy classes of tangle diagrams by the ideal generated by the bracket relation and the relation that the loop value is 6. In this way, every n-string tangle diagram can be written as a linear combination of elements in the basis of TL,. A matrix representation of the inner

(v:'~)@" -+ (v:l2)gn

via the association

of n, and U to the maximal and minimal points respectively, and

132 product (.,

THE CLASSICAL D QUANTUM AN 6j-SYMBOLS
is given as the collection

THE QUANTUM TRACEN D COLOR A REPRESENTATIONS 133
In [21] KOand Smolinski show that all the roots of the equation det T, = 0 are of the form 6 = 2 cos kn/(m and 1 5 k

0 )

+ 1) where 1 < m < n

where w j and wk range over the set of normal form monomials. For example, when n = 3, the matrix is the matrix
- S 3 S2 S2 S2 6 S2 S3 6 6 6 - .

< m. So T, is non-singular unless A is a 4rth root of unity and n > r - 1. This completes the proof. 4.2.2 Remark. The singularity of the pairing (., .) when n >
r - 1 is a key ingredient in Lickerish's [23] construction of the Reshetikhin-Turaev invariants. KO and Smolinsky give an elementary combinatorial proof of this singularity via a recursive construction of the Temperley-Lieb basis. The proof of singularity is originally due to Jones [13]. Our proof of faithfulness breaks down in case the matrix Tn is singular, but according to [7] the representation BA is faithful in a.11 ca.ses except possibly when A = -1. Their proof follows along
the sa.nic li~icsas ours ill t,l~c cla.ssical ca.scl n.llic.11 ; ~ i s o c.o\.cr:, ;.!lo

63 S2 S2 6 .S2 6'

6 d2 S2 6 S3
S2 S2 S3 S3 hl, h2, hlh2, h2hl).

where S is the value associated to any simple closed curve, and the ordered basis of the algebra TL3 is
(13,

because this is the loop value associated to the identification
U

case of A = -1.
4.3 Color representations.

n

I +

[0, iA, - i ~ - ' , 0It. [0, iA, - i ~ - l , 01.

We keep to the case that A is a

More generally, 6 is the loop value in C of the Temperley-Lieb algebra regardless of the representation.
4.2.1 Proof of Theorem 3.3.4. This proof was indicated to

primitive 4rth root of unity with the integer r 2 3. In this case
[T]

= 0. For such values of A define the set of colors to be the

set {0,1/2,1,3/2, us by Paul Melvin. Let T, denote the matrix of the inner product

...,( r - 2)/2).

We will consider representations

vi where j is a color, for these representations are irreducible and
by computing modulo the trace zero representations, we will still have a Clebsch-Gordan theory.
4.3.1 Lemma.

(., .) defined on n-string tangles with respect to the basis of normal
form monomials in the generators 1, h l , . . . ,h,-l. given representation 6 = -A2 - A-2. The representation, Observe that det T, is a polynomial function of thk loop value 6. Under the

The representation V i is not of trace 0 if a is

a color. I t is of trace 0 if a = ( r - 1)/2.
Proof. A computation with weight vectors shows that the rep-

resentation V i is irreducible when a of the Temperley-Lieb algebra into a matrix algebra is faithful when the matrix Tn is non-singular. fices to compute trq(+&) = [2a non-zero unless a = (T - 1)/2.

<

( r - 1)/2. By Schur's

Lemma, any Uq(s1(2)) map is a multiple of the identity. It suf-

+ 1 (Lemma 4.1.4); this trace is 1

134

THE CLASSICAL D QUANTUM AN 6j-SYMBOLS

T H EQUANTUM TRACE AND COLOR REPRESENTATIONS135

4.3.2 Definition.

A q-admissible triple is a triple (a, b, j ) of col-

ors such that
1. a

M 8 v;/' be a Uq(s1(2)) invariant map. Define an invariant map 4 : M -t M by the formula
Proof. Let
+

4 : M 8 v;/'

+ b + j is an integer; + b - j , b + j - a, and j $ a - b are all > 0;
mapping

4= (

1 8~ n) ( 4 8

I) 0 (IM 8 u)

2. a

The role of the last condition is as follows. When A is a primitive 4rth root of unity and when (a, b, j) is q-admissible, then the map

An algebraic computation shows that trq(4) = -trq($). Since M has trace zero, trq(+) = 0, and so trq(+) = 0, which completes the proof.
4.3.5 Lemma. For colors a and 6 , there is a subrepresentation

-

y

ab

is non-zero (Theorem 3.6.6). Consider the computaj

tion of the closed 0-network, @(a,b, j), given in Theorem 3.7.1. In the root of unity case, @(a,b, j ) = 0 when w The inter~retation tha.t t.he corllpositioll is because
"
ab

1

, . I

+ b + .j ~ b> r. - 2.
, ,

o
ab

\/

Ua,b of

6 V: such that 3

va.uislles
1. Ua,b has trace zero, and

y

I j

maps into a summand of ( v ' / ~ ) @ ~ ( ~ + ~ ) of that is
j

trace zero.

2. V j 8 I UaC $ vj) where the sum is over all colors j such that (a, b, j) is a q-admissible triple.
Proof. The proof is by induction on b. The case b = 0 is trivial.

~1

(ej

4.3.3 Lemma. For colors b,

v ;

€4 v;I2

2 2 2

if b = 0; i f 1 1 2 5 b 5 ( r -2)/2.
-.

The case b = 112 is Lemma 4.3.3. Now suppose that 1 5 b 5 ( r - 2)/2. On the one hand, the inductive .hypothesis applied to b - 1 shows that

Observe that

vF1l2has trace zero if b = ( r - 2)/2.

Proof. The relevant maps

j = b f 112; so the result follows by a dimension count.
c 2

y Y2

vj 8 v b-112 ,
2 !

are defined and non-zero

vj 8 (v;

@ vj-l)

v ;

8v ;

$
j:(a,b-1 j)g-admissible

4.3.4 Lemma.

I M is a finite dimensional Uq(s1(2)) repres f

tation of trace zero, then M 8

~ 1also has trace zero. '~

where Ua,b-l has trace zero.

138

T H ECLASSICAL D QUANTUM AN 6j-SYMBOLS

THE QUANTUM TRACE N D COLOR A REPRESENTATIONS 139
is a Uq(sE(2)) endomorphism of (v;'~)@~("'~). Its quantum trace is the value @(a,b, j )

where the sum is over all colors j such that (a, b, j ) is a q-admissible triple, and the summand U' is a subrepresentation of trace 0 . Proof. First we observe that in the case of colors k the JonesWentzl projectors

# 0 (See Remark 3.7.4).

Let W $ U denote

a direct sum decomposition of (v:'~)@~("+~) where W is a subrepresentation complementary to U, and let pw and pu denote the projections onto the indicated summands. Now we compute

+& are defined.

Thus
$j

is defined, for
: Vi + (VA )

(a, b, j ) a q-admissible triple, and the map defined.

112

82j

is

Let V j @ V = W' $ U' be a direct sum dec~mposition ; where W' is a direct sum of color representations and U' is a subrepresentation of trace 0. By Lemma 4.3.5, we just have to prove that under the stated conditions (pa @ pa)( contained in U'. Let U1' = da@db(U1) ( l ~ i ' ~ ) @Since+ ~@ d b ( ~ i @ ~ l ) c ~ ~ +, ~ . is

y

ab

(4j(Vi))) is not
j

The first term is 0 because the image of f is assumed to be in U ; the next two terms are 0 because in a block matrix representation of these maps, the two diagonal blocks are 0. The last term is
0 because it equals the quantum tra.ce of the l1,(s1(2)) inva.ria.nt

for any color t, the multiplication map peld,(v;) is an isomorphism inverse to 4e. So (pa @ ~b)l~,@+,(v;gv,)is an isomorphism that is the inverse of

4,

@

4b.

Now the image of

4.4 The q u a n t u m 6j-symbol - root of unity case. In order lo define the 6j-symbol in the case that A is a primitive 4rth root of unity, we will need to decompose the tensor product

of three color representations into a direct sum of colors and a summand that is of trace 0. We will mod out by the trace 0 piece and construct bases for invariant maps into the quotient. The 6j-symbol, then, will be defined as a change of basis matrix in a family of maps from a color representation to the quotient of the tensor product by a maximal trace 0 summand. Let us turn to the construction. Let A denote a primitive 4rth root of unity, and let a, b, c E (0,112,. . .,( r - 2)/2) - the set of colors.

pose it were. Consider the composition f =

4 4 1 Proposition. ..

140

THE CLASSICALD QUANTUM AN 6j-SYMBOLS
= Pa 8 p 8 pc. T h e n b

THE QUANTUM TRACEN D COLOR A REPRESENTATIONS 141

Let

pabc

n:(a,b,n) is q-ad. = n:(a,b,n) is q-ad.

ab

v,.

8 v 8 v, = ; .

(y

8 :

1)

o(+n~+c)(v~8vj)@(u8vA")

where Ul and Uz are trace 0 subrepresentations of V i @ V: @ V j .
Proof. \,\:e lla.\:c i s v ~ ~ ~ v r p l l(*)~ ~ ~ s is

-

n,k:(a,b,n)&(n,c,k) are q-ad.

€3

(v;

8v ; )

8 vi

=

@
{n:(a,b,n) is q-ad.}

v,"@Vj@(Ua,b@V;)

By computation (*) and the Remak-Krull-Schmidt Theorem, the subrepresentation Ul must be of trace 0. This completes the

N

-

{n:(a,b,n) is q-ad.}
-

(

vjaun,c

{k:(n,c,k) is q-ad.)

)

@(ua,b@vi)

proof of the first equality. Now we turn to establish the second equality. We claim that up to isomorphism,

{n,k:(a,b,n)&(n,c,k) are q-admis.} @

@

v,k

(

@

un,c@(ua,b@vi)

{n:(a,b,n) is q-ad.}

)

v,.
On

8 v 63 vz; 2 ; l

(

{~,k:(bc,~)&(a,~,k)are q admis.)

@

v-

)

@

ut

where the term @{n:(a,b,n) is q-admis.) Un,c @ (Ua,b 8 V;) right of the last equality is of trace 0. Furthermore,

the

where U' has trace 0. By the computation (*) and the RemakKrull-Schmidt Theorem, it suffices to show (*)

(v,"8 v;)

8 vi

{~,k:(b,c,~)&(a,~,k)are Q admis.)

03

k v* 2

{n,k:(a,b,n)&(n,c,k)areq admis.)

@

V" A

THE CLASSICAL D QUANTUM - S Y M B O L S AN 6j
be a U,(s1(2)) invariant map. Then there are unioue comvlex num.bers d, and map S : VJ
-+ V j

145 T H EQUANTUM TRACE N D COLOR A REPRESENTATIONS
maw. Thus S = S', and by the linear independence of the maps

8 V 8 V , such that ;

{ n : ( a ,b, n ) & ( n ,c, k) are q-admis.) and

4.4.5 Definition.

In light of the preceding Lemma, we can

define in the case that A is a primitive 4rth root of unity the 6 j I

symbol to be the coefficient

s ( v ~ )contained in a trace zero summand of V j 8 V is ;

8 V,L.
tion:

1:

\

}

in the following epuaq

Proof. The existence of the numbers d, and map S is immediate
from Proposition 4.4.1. Suppose that dk and S' were another
solution. so that

where S maps into a summand of trace 0, pat,, = pa € pi, 8 p,, 3 and the sum is taken over the set of colors n such that the triples

( a ,b, n ) and ( n , c, k) are q-admissible.
4.4.6 Theorem.

~ e At denote a primitive 4rth root of unity.

Then the quantum 6j-symbols satisfy the Elliott-Biedenharn and
By the preceding lemma, the image of the right side S' - S lies within any maximal trace zero summand of V j 8 V @ V,L, and ; by Proposition 4.4.1 the image of the left side must intersect such a summand trivially. Therefore both sides are the zero

orthogonality identities as stated in Theorems 3.8.6 and 3.8.5 where the sums are over q-admissible indices. Moreover for such values of A, the normalized 6j-symbols are defined and the identities stated in Theorem 3.1 1.4 hold.

c

k

j

9

T H ECLASSICAL D QUANTUM AN 6j-SYMBOLS

149 T H EQUANTUM TRACE AND COLOR REPRESENTATIONS

We form a closed network by reflecting from which it follows that

the horizontal axis to obtain the diagram

a

b

c

This is juxtaposed (with the upper index n changed to s ) to

Now we establish the Elliott-Biedenharn identity using the same technique. As in the proof of Theorem 2.6.7, we have the identity -

(.he top os

I

1;

t,o C ~ I . ~ It.l~c! I I C ~ I V C , I . I < : L

where T is the tree depicted on the left below - that holds in the

N = N(a, b, c, n, j, k) =

a

sense of insertion into closed networks.

Then the network, N , is closed by joining the arcs that are labeled by k. Let N denote this closed network, or its value in C . We Let T* = T*(a, b, c, d, s, t, g) denote the mirror image of T through a horizontal axis as shown on the right of the diagram above. Consider the closed network Closure(T*T).

150

T H ECLASSICALD QUANTUM AN 6j-SYMBOLS

5

The Turaev-Viro Invariant

In this last section we explain how to use the normalized 6j-symbol (in the case that A is a primitive 4rth root of unity) to give the definition of the Turaev-Viro invariants of 3-dimensional manifolds. Computations of this invariant can be found in the paper [32] and the book [18]. Specialization to the root of unity case is necessary so that the sum in the definition of the invariant is a finite sum (physicist's renormalization). While the invariants have not distinguished 3-manifolds that We compute Closure(T*T) = Sh,tO(t, d, c)Ss,kO(t,b, s)O(a, s, g)/(AtAs). cannot be distinguished in other ways, new applications of these invariants are expected. Furthermore, the framework of a topological quantum field theory - into which the Turaev-Viro invariants fit - is quite general, and is currently being explored in its own right. In particular, there is hope that interesting Cdimensional
g c ~ ~ ~ l a l i ~ a t calln Ijt. foulld. n ~ r dt h a t thc D o ~ l a l d s o nirir.;lr.ia~~t<. io s

for example, can be defined as state summations [4]. We mention the following interesting problem: Suppose two manifolds of dimension 3 have the same Turaev-Viro invariants, in what ways are they similar? In other words, what qualitative features do the Turaev-Viro invariants distinguish?

.@(t,d, c)@(t,b, s)@(a,s, 9)/(AtAs).
It follows that for any choice of s and t so that (t,d, c), (t, b,s), and (a, s, g) are q-admissible: 5 1 Definition. . Fix an integer r 2 3, let A = eTil(2T), and let q = A2. Let M denote a triangulated 3-dimensional closed manifold. Let t denote the number of vertices, let {El,. ..,E,) denote the set of edges, and let {TI,. . .,T,) denote the set of tetrahedra of the triangulation. Let C = (0,112,. . . ,( r - 2)/2) denote the set of colors associated to the integer r. A coloring of

M is a mapping f : {El,. ..,E,)
This completes the proof.
C I

+ C. An admissible coloring is

a coloring such that for each triangle with edges Ee, Em, and En the triple (f (El), f(E,), f (E,)) is a q-admissible triple of colors.

152

THECLASSICAL QUANTUM AND 6j-SYMBOLS
on the left can be manipulated to the figure on the right and vice versa.

Suppose that a coloring is admissible, and consider a tetrahedron T with colors a, b, c, j, k, n associated to its edges so that the triples (a, b, n), ( n , ~k), (a, j, k), and (b, c, j ) are admissible, and , these are the labels on the bounding triangles of the tetrahedron.
r
I

Associate the symbol T~ =

I: : ;1
7L

to this tetrahedron.
0

The value associated to the coloring the quantity.

1 of the 3-manifold M is
?,I

where

A = AT'
3

{k,e:(j,k,l)isq-admis.) It is a consequence of the orthogonality identity that the quantity

C

AkAl.

(I;,~:(~,rl-,l)is q-admis.} is independent of j (see [32, 181 for a proof). The T~raev-Viroinvariant of the 3-manifold M is the state sum

' 4

\y

In the first figure two tetrahedra with edges (a, b,e,g,h, k) and (,-, d, e, f , g, h ) are glued along their common triangular face (g, h, e). Then an edge labelled j is inserted, so the polyhedron

IMI =
f

IMI

f

is the union of 3 tetrahedra: (a, b, e, e , f, j), (b, e, di h, j, li), and (,, d , f , g, j, k). The Elliott-Biedenharn identity

where the sum ranges over a l admissible colorings f of the given l triangulation.
5.1.1 Theorem (Turaev-Viro [32]). The value IMI E C is independent of the triangulation chosen; as such it is an invariant

[;::I., . [ R"1,
b
e

of the manifold M.

jl

[ j d k 1

d h]

Proof. The Pachner Theorem for triangulations of $manifolds,
states that any two triangulations of a given 3-manifold are related by a sequence of the two moves depicted below where the figure

156

THE CLASSICAL QUANTUM AND 6j-SYMBOLS
In Kauffman-Lins

5.1.2 Definition (Kauffman-Lins 1 8 ) 11.

the following alternative definition of the Turaev-Viro invariant is presented. First a triangulation of a 3-manifold is chosen and dualized. The vertices of the dual correspond to the tetrahedra of the triangulation, the edges in the dual correspond to the faces of the triangulation, and the faces in the dual correspond to edges in the triangulation. Colors are associated to the faces of the dual in such a way that three colors coincident to an edge form a qadmissible triple. To a vertex, at which six faces meet, the value of a tetrahedral spin network is associated. The spins on the edges of the tetrahedral network are the colors associated to the faces of the dual, and the edges of the dual correspond to the vertices of the tetrahedral network. To an edge in the dual with colors a , b and 7 coincident the value @ ( a , , ~ is associated. Since these b ) for111a q-admissible triple, the value of 0 is defined and where j is the color on the face. Thus a state, S , of the triangulation is an assignment of colors
. ..
.,.
1s

f

If a given edge, e, forms a simple closed curve with no vertices from the dual complex, then let ~ ( e = 0; if the edge has a vertex, ) then let ~ ( e= 1. With these conventions, a state sum is defined ) by the formula

The proof that this does not depend on the choice of triangulations is given in [18]; the proof depends on expressing the tetrahedral networks in terms of the 6j-symbol. In Kauffman-Lins a nice glimpse of the shadow world is presented, as well.
5.1.3 Theorein (Piunikhin [ZB]).
'

The Iicr~<ffnza~n-Liizs (1e.b-

non-

~zitio~zi ~ the Y L I . U ~ lfiro definitioizs coi~lcide. u t U-

zero. To each colored face in the dual we associate the value A,

Proof. This is a computation dependent on the definition of the
r
!

to the "faces7'of the dual such that the colors coincident at an edge form a q-admissible triple. (The term "face" is in quotes above because as the dual complex to the triangulation is deformed, some of the 2-dimensional pieces may not be 2-.cells. For example, the Matveev bubble move introduces an annular face.) To such a state, we have TET(v1S) the value of the tetrahedral spin net associated to each a vertex v dual to the given cell. To an edge, E , with colors (a, b, j ) coincident, we associate the value O(EIS) = @(a,b, j). And to a face, f , with color j ( f ) = j ( f IS), associate ) the value Aj(f). Let ~ ( f denote the Euler characteristic of the 2-dimensional face f where the term face is interpreted as above.

In the current work we have presented the diagrammatics of the classical and quantum representation theory
with topological applications in mind. One major focus has been the Clebsch-Gordan theory and the explicit construction of the Gj-symbol in all three cases - classical, generic quantum, and quantum root of unity. In the process of developing this theory, we have touched on some other topological aspects that deserve mentioning. The Jones polynomial [12], which is an invariant of knotted and linked curves in 3-dimensional space, is the starting point of the quantum topology invariants. The bracket identity leads

5.2 Epilogue.

158

THE CLASSICAL AND

Q U A N T U M 63-SYMBOLS

THE TURAEV-VIRO INVARIANTS

159

directly to a definition, and this construction can be found in [16], for example. The quotient of the quantum group Uq(s1(2)), when A is a primitive 4rth root of unity, by relations Er = Fr = 0 and K4' =
1 has the structure of a modular ribbon Hopf algebra as defined

known algebraic structures. We have seen this happen already in the passage from classical sl(2) to quantum sZ(2). For computations of the Turaev-Viro invariants see [18]. Finally, there are deep connections to theoretical physics that require much further study from the mathematical, theoretical, and experimental sides. The mathematical aspect that is the most problematic is the definition of functional integration - which might be thought of as a continuous version of the state sum method. A rigorous definition of the functional integral will lead to analytic interpretations of the algebra and of the topology, and
l such intepretations will certainly shed light on al of the aspects

by Reshetikhin-Turaev [29]. Rather than explicitly describing this structure we have worked with the algebra via its representations. However, in doing so, we have verified that the ribbon structure is present. The modular ribbon Hopf algebra gives the ReshetikhinTuraev invariant, and Lickorish [23] has presented the definition of the invariant in a diagrammatic form. From the present point of view, there is not much more that needs to be done to get to that formulation. The details of that construction can also be found in [161.

of the theory.

T h e Turaev-Viro illvariant is an esanlple ol a topologica.1quantum field theory (TQFT), namely a functor from the category of smooth manifold cobordisms to the category of Hilbert spaces. A great deal of effort is currently being exerted towards finding new extended topological field theories, and towards finding higher dimensional analogues. One formulation of the higher dimensional theories is found in Lawrence [22]; from that point of view the structure that is associated to a 3-manifold is a "3-algebra," and the quantum 6j-symbol gives an explicit construction of such an algebra. In dimension 4, an example of a 4-algebra would give rise to a state sum invariant of the type constructed here. There are very good reasons for searching for such higher dimensional invariants. Evidence for their existence is given by the solutions to the tetrahedral equation [35] which is an analogue of the Yang-Baxter equation. Furthermore, non-trivial examples of these higher algebraic structures will give new and in-depth meaning to the well-

References
[I] Alexander, J.W., The Combinatorial Theory of Complexes, Annals of Math. 31 (1930), 175-186. [2] Biedenharn, L. C. and Louck, J . D., "The Racah-Wigner Algebra in Quantum Theory," Encyclopedia of Mathematics, Addison-Wesley (1981). [3] Brink, D. M. and Satchler, G. R., "Angular Momentum," Oxford University Press 2nd edition, (1975). [4] Crane, L. and Frenkel, Igor, Four Dimensional Topological Quantum Field Theory, Hopf Categories, and Canonical Basis, J. Math. Phys. 35 (lo), (Oct 1994), p. 5136. [5] Drinfel'd, V. G., Quantum Groups, Proc. ICM-86 (Berkeley), vol.1, Amer.Math.Soc., (1987), 798-820. [6] Drinfe'ld,

Reprinted in Jimbo, M., "Yang-Baxter Equation in Integrable Systems," World Scientific Publishing Co., (Singapore 1989). [I 1 Jimbo, M., "Yang-Baxter Equation in Integrable Systems," 1 World Scientific Publishing Co., (Singapore 1989). [12] Jones, V. F. R., Hecke Algebra Representations of Braid Groups and Link Polynomials, Ann. of Math. 126 (1987), 335388. Reprinted in Kohno "New Developments in the Theory of Knots," World Scientific Publishing (Singapore 1989). [13] Jones, V. F. R., Index for Subfactors, Inventiones Math. 72 (1983), 1-25. Reprinted in Kohno "New Developments in the Theory of Knots," World Scientific Publishing (Singapore 1989). [14] Kauffman, L., 'Spin Netzoorks and the Jones Polynomial, 'h15tor Newsletter. ?;n 20 (8So\ernl>cl 1959). hlathcnlatlcs Institute, Oxford, 25-30. [15] Kauffman, L., SL(2),-Spin Networks, Twistor Newsletter, No. 32 (12 March 1991), Mathematics Institute, Oxford, 1014. [16] Kauffman, L., "Knots and Physics," World Scientific Publishing (Singapore 1991). 1171 Kauffman, L., Map Coloring, q-Deformed Spin Networks, and the Tureav- Viro Invariants for PManifolds, International Journal of Modern Physics B, Vol. 6, Nos. 11 & 12, (1992) 1765-1794. 1181 Kauffman, L. and Lins, S., "The Temperley-Lieb Algebra Recoupling Theory and Invariants of 3-Manifolds," Annals of

V. G.,

Quasi-Hopf Algebras and linizhnik-

Zamolodchikov Equations, Research Reports in Physics, Problems of Modern Quantum Field Theory, (Circa 1990). [7] Goodman, F. M. and Wentzl, H., The Temperley-Lieb Algebra at Roots of Unity, Pacific Journal of Math. Vol 161, No.2 (1993), 307-334. [8] Humphreys, J. E., "Introduction to Lie Algebras and Representation Theory," Springer-Verlag (New York 1972). [9] Jacobson, N., "Basic Algebra 11," W. H. Freeman Co. (San Francisco 1980). [lo] Jimbo, M., A q-Digerence Analogue of U(g) and the YangBaxter Equation, Letters Math. Phys. 10 (1985), 63-69. 160