Chapter 70

Compact quantum groups

70.1 Definitions
Literature: [558, 559, 560, 561] and Wikipedia: quantum group and compact quantum group.
For introductions about Hopf algebra and some related topics such as quantum groups and defor-
mation quantization of (co)-Poisson structures, see [480, 562].

Définition 70.1.
A compact quantum group is a pair G “ pA, q where A is an unital separable C ˚ -algebra and
: A Ñ A b A is a ˚-homomorphism such that
(1) pId b Idq “ pId b q ;
(2) The sets
tpb b q pcq tel que b, c P Au,
(70.1)
tp b bq pcq tel que b, c P Au
are linearly dense in A b A.

70.1.1 Example: representation of groups

If u is a finite-dimensional representation of a group G, we can build the ˚-subalgebra of CpGq
generated by the elements uij and –puij q where – is defined by

–puij qpgq “ uij pg ´1 q. (70.2)

Using the definition 69.113, we turn this algebra into a Hopf algebra 1 . Let us particularize the
definitions. For the coproduct we have
ÿ
puij q “ uik b ukj (70.3)
k

because if g and h are elements of G, the coproduct on CpGq is equal to

` ˘ ` ˘ ÿ ´ÿ ¯
puij qpg, hq “ uij pghq “ upghq ij “ upgquphq ij “ upgqik uphqkj “ uik b ukj pg, hq
k k
(70.4)
where we used the fact that u is a representation and the fact that, by definition, uij pgq “ upgqij .
The counit is given by
‘puij q “ ”ij . (70.5)
since the neutral element e is always represented by the unit matrix. The antipode is

–puij qpgq “ uij pg ´1 q. (70.6)

1. This is not trivial since nothing guarantee a priori that the right hand sides of the definitions are elements of
our small algebra.

2735
2736 CHAPTER 70. COMPACT QUANTUM GROUPS

The unit is given by the constant function 1. Fortunately, it turns out that this function is a
combination of the functions uij :
ÿ ÿ
1“ u1k –puk1 q “ –pu1k quk1 . (70.7)
k k
Indeed, ÿ ÿ
u1k pgq–puk1 qpgq “ u1k pgquk1 pg ´1 q
k
´k ¯
“ upgqupg ´1 q (70.8)
11
“ upeq11
“ 1.
Let us check one of the diagrams (52.44):
‘bId
b CpGq o CpGq b CpGq (70.9)
g O

Â
CpGq
We have ÿ
p‘ b Idq puij q “ p‘ b Idq uik b ukj
k
ÿ
“ ”ik b ukj
k
ÿ (70.10)
“ 1 b ”ik ukj
k
“ 1 b uij
“ Âpuij q.

70.1.2 Matrix quantum group

Définition 70.2.
A matrix quantum group is a pair pC, uq where C is a C ˚ -algebra and u “ puij q is a matrix
with entries in C such that
(1) The ˚-subalgebra C0 generated by the elements uij is dense in C.
(2) There exists a C ˚ -algebra homomorphism : C Ñ C b C such that
ÿ
puij q “ uik b ukj . (70.11)
k

(3) There exists an antimultiplicative map – : C0 Ñ C0 called coinverse such that

` ˘
(a) – –pv ˚ q˚ “ v for any v P C0 ,
∞ ∞
(b) k –puik qukj “ k uik –pukj q “ ”ij where is the unit in C.

70.1.2.1 Example: SUq p2q

As an example, we give the quantum matrix group SUq p2q. Let q be a positive real number.
As C ˚ -algebra, SUq p2q is generated by the elements a and b and the relations 2
ba “ qab
b a “ qab˚
˚

bb˚ “ b˚ b (70.12)
2 ˚
a a`q b b“1
˚

aa˚ ` bb˚ “ 1
2. We follow the notations of [563].
70.1. DEFINITIONS 2737
´ ` ˘ ¯
We build a compact matrix quantum group C SUq p2q , u where
ˆ ˙
a qb
u“ . (70.13)
´b˚ a˚
` ˘
The Hopf-˚-algebra structure on C SUq p2q is given by the relations in the definition 70.2. We
have to check that these are compatible.
We have
ÿ
paq “ pu11 q “ u1k b uk1 “ u11 b u11 ` u12 b u21 “ a b a ´ qb b b˚ , (70.14)
k

and
ÿ
pa˚ q “ pu22 q “ u2k b uk2 “ u21 b u12 ` u22 b u22 “ ´qb˚ b b ` a˚ b a˚ , (70.15)
k

which is compatible with (70.14). Doing the same with b “ 1q u12 (and checking the compatibility
with b˚ ), we find the coproduct
paq “ a b a ´ qb b b˚
(70.16)
pbq “ b b a˚ ` a b b.
The counit ‘puij q “ ”ij provides
‘paq “ 1
(70.17)
‘pbq “ 0.
Using the relations (70.12), we can check that the condition (3) of definition 70.2 are satisfied
by the antipode
–paq “ a˚ , –pa˚ q “ a,
(70.18)
–pbq “ ´qb, –pb˚ q “ ´q ´1 b˚ .
As an example,
` ˘ ` 1 ˘ 1
– –pb˚ q˚ “ – p´ b˚ q˚ “ ´ –pbq “ b (70.19)
q q
∞
and k –puik quk2 “ ”12 “ 0 since

–pu11 qu12 ` –pu12 qu22 “ –paqqb ` q–pbqa˚ “ qa˚ b ´ q 2 ba˚ “ 0 (70.20)

because a˚ b “ pb˚ aq˚ “ pqab˚ q˚ “ qba˚ . `

All this structure provides a Hopf ˚-algebra named C SUq p2qq.

70.1.2.2 Corepresentation
A corepresentation of the quantum group pA, q is an element u P n pAq such that
n
ÿ
pukl q “ ukj b ujl (70.21)
j“1

for every k, l P t1, . . . nu.

70.1.3 Haar measure on compact quantum groups

Let A be the C ˚ -algebra CpGq where G is a compact group. The Haar measure µ on G is
the unique regular Borel measure on G such that µpGq “ 1 and
ª ª ª
f pstqdµpsq “ f ptsqdµpsq “ f psqdµpsq. (70.22)
G G G
2738 CHAPTER 70. COMPACT QUANTUM GROUPS

We can see the Haar measure µ as a positive functional Ï : A Ñ ,

ª
Ïpf q “ f psqdµpsq. (70.23)
G

If h P A is such that phq “ hp1q b hp2q (sum implied), then we have

` ˘
phqps, tq “ hpstq “ hp1q b hp2q ps, tq “ hp1q psqhp2q ptq. (70.24)

Thus we have
ª ª
` ˘
phqps, tqdµpsq “ hp1q psqhp2q ptqdµpsq “ pÏ b Idq hp1q b hp2q ptq “ pÏ b Idq phq. (70.25)
G G

Using the invariance of the Haar measure, this is also equal to

ª ª
phqps, tqdµpsq “ hpstqdµpsq
G ªG
“ hptsqdµpsq
ªG (70.26)
“ hp1q ptqhp2q psqdµpsq
G
` ˘
“ pId bÏq hp1q b hp2q ptq
“ pId bÏq phq.

Since CpGq b CCpGq is dense in CpG ˆ Gq, we have

pId bÏq phq “ pÏ b Idq phq “ Ïphq1 (70.27)

for every h P A.
The generalisation of that result to an arbitrary compact quantum group is the following
theorem.

Théorème 70.3.
If pA, q is a compact quantum group, there exists one and only one state „ : A Ñ such that

pId b„q paq “ „paq1 “ p„ b Idq paq (70.28)

for every a P A.

The unique state guaranteed by that theorem is called the Haar state of A. If Ê and Ê 1 are
linear functionals on A, we define the convolution by

Ê ˚ a “ pId bÊq paq

a ˚ Ê “ pÊ b Idq paq (70.29)
1 1
Ê ˚ Ê “ pÊ b Ê q paq.

The first two are elements of A when the lase one is a linear functional on A. Remember that here
the map plays the role of a coproduct; when the context is about a coalgebra, we need to adapt
the notation and write instead of in the definitions 70.29.
The Haar state is the state „ such that

„ ˚ a “ a ˚ „ “ „paq1. (70.30)
70.2. CONSTRUCTION OF SUQ pN q 2739

70.2 Construction of SUq pnq

In order to build the quantum group SUq pnq, we will proceed the following step[560]
(1) we consider the n2 elements U “ puij q and we name Aq the algebra generated by these
elements;
(2) we impose some relations;
(3) we define the determinant D;
(4) we impose D “ 1;
(5) we define the involution uij fiÑ u˚ij ;
(6) we impose U U ˚ “ U ˚ U “ 1.
Notice that the point (5) will not introduce new elements u˚ij . It will define u˚ij as combination of
the uij ’s.
The first relations that we impose are
ÿ ÿ mp
kl
Rij ukm ulp “ Rkl uik ujl (70.31)
kl kl

where R P n2 p q is the matrix given by

ii
Rii “ q ´1 ji
Rij “ 1 if i ‰ j
ij
Rij “ q ´1 ´ q if i ° j (70.32)
kl
Rij “ 0 otherwise.

Remarque 70.4.
In [564], there is an alternative compact form for the same constraints.

Proposition 70.5.

uij uil “ quil uij if j † l (70.33a)

uij ukj “ qukj uij if i † k (70.33b)
uij ukl “ ukl uij if i ° k, j † l (70.33c)
uij ukl ´ ukl uij “ pq ´ q ´1
quil ukj if i † k, j † l. (70.33d)

Proof. Let us check the relation (70.33c). If we consider i, j, m, p with i † j and m † p, the only
non vanishing Rij
kl is k “ j, l “ i and the only non vanishing Rmp is k “ p, l “ m. Thus we
kl
immediately get ujm uip “ uip ujm .

A particular consequence of (70.33c) is that

uij uji “ uji uij ; (70.34)

an element commutes with its “transposed”.

Lemme 70.6.
If l “ minpi, jq with i ‰ j then
uij ull “ q ´1 ull uij . (70.35)

Proof. If l “ i then l † j and the relation (70.33a) provides ulj ull “ q ´1 ull ulj . Now if l “ j, then
l † i and the relation (70.33b) provides uil ull “ q ´1 ull uil .

Lemme 70.7.
If l “ maxpi, jq with i ‰ j then
uij ull “ qukk uij . (70.36)
2740 CHAPTER 70. COMPACT QUANTUM GROUPS

Proof. If l “ i, we have j † i and the relation (70.33a) leads to uli ull “ qull uli . If l “ j, then i † j
and the relation (70.33b) shows uil ull “ qull uil .

The R matrix is a Yang-Baxter operator:

Rp12q Rp23q Rp12q “ Rp23q Rp12q Rp23q (70.37)

where the operators Rp12q , Rp23q : b b Ñ b b are defined by

Rp12q “ R b Id
(70.38)
Rp23q “ Id bR.
Now we define the homomorphisms
: Aq Ñ Aq b Aq
n
ÿ (70.39)
puij q ބ uik b ukj
k“1

and
‘ : Aq Ñ
(70.40)
‘puij q “ ”ij .
We also set ‘p1q “ 1 as required for a counit (definition 52.23). At this point Aq is a bialgebra.

70.2.1 Determinant
We introduce the quantum determinant
ÿ
D“ p´qq|‡| u1‡p1q . . . un‡pnq (70.41)
‡PSn

where |‡| is the length of the permutation. We also introduce the quantum minor Dij
ÿ
Dij “ p´qq|‡| u1‡p1q . . . uz
i‡piq . . . un‡pnq . (70.42)
‡ : t1,...,ı̂,...,nuÑt1,...,ä̂,...,nu

Let I and J be subsets of size t of t1, . . . , nu. We introduce the quantum minor
ÿ
rI|Js “ p´qq|‡| ui‡p1q,j1 . . . ui‡ptq,jt . (70.43)
‡PSt

The coproduct of that minor has the simple expression[565]

Proposition 70.8.
We have ÿ
rI|Js “ rI|Ks b rK|Is (70.44)
|K|“|I|

where the sum is over all the subset K Ñ t1, . . . , nu of size equal to the size of I.

Sketch of the proof. We prove by induction on the size of I. Using the fact that is an homo-
morphism,
ÿ n
ÿ ` ˘
rI|Js “ p´qq|‡| ui‡p1q k b ukj1 ui‡p2q j2 . . . ui‡pnq jn
‡PSt k“1
ÿt ÿ n
ÿ ` ˘
“ p´qq|‡| ulk b ukj1 ui‡p2q j2 . . . ui‡pnq jn (70.45)
l“1 ‡PSt plq k“1
t
ÿ ÿ n
ÿ ÿ
“ p´qq|‡| ulk b ukj1 rIztlu|KsrK|Jztj1 us
l“1 ‡PSt plq k“1 |K|“|I|´1
70.2. CONSTRUCTION OF SUQ pN q 2741

where St plq is the subset of St of permutations such that ‡p1q “ l.

Problèmes et choses à faire

This proof is not finished. I think that there remain a few combinatorial work in order to get the result. At least it looks very like everything is

going to sum up correctly.

Using lemma 70.8, we prove that

pDq “ D b D. (70.46)
We also have ‘pDq “ 1.
Now we add to Aq an element called D´1 on which we impose the conditions
DD´1 “ D´1 D “ 1. (70.47)
We extend and ‘ by
pD´1 q “ D´1 b D´1
(70.48)
‘pD´1 q “ 1.
` ˘ ` ˘
The extension of Aq by D´1 is Pol Uq pnq , or Pol SUq pnq if we impose the extra relation
D “ 1.
For the involution we define Ÿ by
Ÿpuij q “ p´qqi´j Dji D´1
(70.49)
ŸpD´1 q “ D.
` ˘ ` ˘ ` ˘
This extends in an unique way to Ÿ : Pol Uq pnq Ñ Pol Uq pnq in such a way that Pol Uq pnq
becomes a Hopf algebra with Ÿ as antipode.
We introduce the involution by
u˚ij “ Ÿpuji q
(70.50)
pD´1 q˚ “ D.
We have DD˚ “ 1 “ D˚ D.

Lemme 70.9.
We have ‘pu˚ij q “ ”ij .

Proof. If we apply ‘ on the definition (70.42) we get a non vanishing contribution only from the
term ‡ “ Id. Thus we see that ‘pDij q “ ”ij and the result follows immediately.
Notice that this result is part of the fact that PolpUq pnqq is an Hopf algebra with involution
(see lemma 52.34).
Using the formula
ÿn
”ij D “ p´qqk´j uik Djk , (70.51)
k“1
one proves that
n
ÿ n
ÿ
uik u˚jk “ ”ij 1 “ u˚ki ukj . (70.52)
k“1 k“1

Lemme 70.10.
In SUq pnq we have the relations[560, 564]
u˚kl uij “ uij u˚kl if k ‰ i (70.53a)
ÿ
u˚ij ukj “ q ´1 ukj u˚ij ` pq ´1 ´ qq uks u˚is if i ‰ k (70.53b)
s†j
ÿ
u˚ij uik “ quik u˚ij ` pq 2 ´ 1q u˚sj usk if j ‰ k (70.53c)
s°i
ÿ ÿ
u˚ij uij “ uij u˚ij ` pq 2 ´ 1q u˚sj usj ` p1 ´ q 2 q uis u˚is (70.53d)
s°i s†j

where i, j, k, l P t1, 2, . . . , nu and the sums are running up to n.

2742 CHAPTER 70. COMPACT QUANTUM GROUPS

Proof. We are starting from equation (70.31) that we multiply on the left by u˚ir , on the right by
u˚sp and sum over i and p:
ÿ ÿ mp
kl ˚
Rij uir ukm ulp u˚sp “ Rkl u˚ir uik ujl u˚sp . (70.54)
pikl irkl

Using (70.52), the sum over p is easy to perform on the left hand side while the sum over i is easy
to perform on the right hand side:
ÿ ÿ mp
kl ˚
Rij uir ukm ”ls “ Rkl ”rk ujl u˚sp . (70.55)
ikl pkl

For the sake of unifying the notations, we rename the summation variable p to i in the right hand
side: ÿ ÿ
Rijks ˚
uir ukm “ mi
Rrl ujl u˚si . (70.56)
ik il
At this point we have four possibilities following s “ j or s ‰ j and m “ r or not. We look at the
case s “ j, m “ r. In the expression ÿ
mi
Rml usl u˚si , (70.57)
il
only the term i “ l is not zero. Thus the equality reduces to
ÿ ÿ
ks ˚
Rks ukm ukm “ mi
Rmi usi u˚si . (70.58)
k i

We divide the sum over k into k “ s and k ‰ s and the sum over i into i “ m and i ‰ m:
ÿ ÿ
ss ˚
Rss usm usm ` ks ˚
Rks mm
ukm ukm “ Rmm usm u˚sm ` mi
Rmi usi u˚si . (70.59)
j°s i†m

Changing the names s Ñ i and m Ñ j we have

ÿ ÿ
q ´1 u˚ij uij “ uij u˚ij ` pq 2 ´ 1q u˚kj ukj ` p1 ´ q 2 q uik u˚ uik . (70.60)
k°i k†j

This proves the equality (70.53d).

Problèmes et choses à faire

For the other ones, I think that one has to work with the other possibilities about s “ j and m “ r. To be checked.

The quantum group SUq pnq is the unital C ˚ -algebra obtained from the completion of Aq on
which the map extends to a C ˚ -homomorphism
: SUq pnq Ñ SUq pnq b SUq pnq. (70.61)
Problèmes et choses à faire

Is the completion with that requirement unique?

70.2.2 Norm
Proposition 70.11.
If fi is a representation of Aq on D, then }fipuij q}2 § 1.

Proof. Using relation (70.52) we have for every f P D that

}f }2 “ xf, f y
ÿ
“ x fipuik q˚ fipuik qf, f y
k
ÿ
“ xfipuik qf, fipuik qf y (70.62)
k
ÿ
“ }fipuik qf }2 .
k
70.3. REPRESENTATIONS OF SUQ p3q 2743

If }f } “ 1, we thus have }fipuik q} § 1 and

}fipuik q}2 “ sup }fipuik qf }2 § 1. (70.63)

}f }“1

Corollaire 70.12.
The operator Id ´qfipuij q is invertible.
“ ‰
Proof. Since q † 1 and }fipuij q} § 1, it is impossible to have Id ´qfipuij q f “ 0 without violating
}fipuij q} § 1.

70.3 Representations of SUq p3q

We are following [564].

Lemme 70.13.
In SUq p3q we have the relations

u11 u˚11 “ p1 ´ q 2 q ` q 2 u˚11 u11 (70.64a)

u12 u˚12 “ q 2 p1 ´ q 2 q ` q 2 u˚12 u12 ´ q 2 p1 ´ q 2 qu˚11 u11 (70.64b)
2
u21 u˚21 “ p1 ´ q q ` q 2 u˚21 u21 ´ p1 ´ 2
q qu˚11 u11 (70.64c)
2
u22 u˚22 “ p1 ´ q q ` q 2 u˚22 u22 ´ p1 ´ q 2 qpu˚21 u21 ` u˚12 u12 q ´ p1 ´ q 2 q2 u˚31 u31 . (70.64d)

Proof. Let us set i “ j “ 1 in (70.53d). The second sum vanishes while we reform the first sum
by adding and substracting pq 2 ´ 1qu˚11 u11 . What we obtain is
ÿ
u˚11 u11 “ u11 u˚11 ` pq 2 ´ 1q u˚k1 uk1 ´pq 2 ´ 1qu˚11 u11 . (70.65)
loooomoooon
k
“1

So
q 2 u˚11 u11 “ u11 u˚11 ` pq 2 ´ 1q (70.66)
which is the relation (70.64a).
In order to prove the relation (70.64b) we start from

u12 u˚12 “ u˚12 u12 ´ pq 2 ´ 1qpu ˚ ˚ 2 ˚

22 u22 ` u32 u32 q ´ p1 ´ q qu11 u11
loooooooomoooooooon (70.67)
“1´u˚
12 u12

in which we substitute the value of u11 u˚11 given by (70.64a).

Problèmes et choses à faire

I guess this is the same for the other relations, setting other values for i and j.

Lemme 70.14.
In SUq p3q we have the relations

u11 pu˚11 qn “ p1 ´ q 2n qpu˚11 qn´1 ` q 2n pu˚11 qn u11 (70.68a)

u12 pu˚12 qn “ q p1 ´ q qpu˚12 qn´1 ` q 2n pu˚12 qn u12 ´ q 2 p1 ´ q 2n qpu˚12 qn´1 u˚11 u11
2 2n
(70.68b)
u21 pu˚21 qn “ p1 ´ q 2n qpu˚21 qn´1 ` q 2n pu˚21 qn u21 ´ p1 ´ q 2n qpu˚21 qn´1 u˚11 u11 (70.68c)
1 ´ q 2n ˚ n´1 ˚
u22 pu˚12 qn “ q n pu˚12 qn u22 ´ q 2 pu12 q u11 u21 (70.68d)
qn
1 ´ q 2n ˚ n´1 ˚
u22 pu˚21 qn “ q n pu˚21 qn u22 ´ pu21 q u11 u12 (70.68e)
qn
1 ´ q 2n ˚ n´1 ˚
u23 pu˚21 qn “ q n pu˚21 qn u23 ´ pu21 q u11 u13 . (70.68f)
qn
2744 CHAPTER 70. COMPACT QUANTUM GROUPS

Proof. Equation (70.68a) is checked by setting i “ j “ 1 in equation (70.53d).

Let us now check the equation (70.68d). First, writing (70.53b) with i “ j “ l “ 1, k “ 2 and
taking the adjoint produces
u21 u˚11 “ qu˚11 u21 . (70.69)
Now if we write equation (70.53b) with k “ 1 and l “ i “ j “ 2 we have

u22 u˚12 “ qu˚12 u22 ´ p1 ´ q 2 qu21 u˚11 . (70.70)

Substituting (70.69) into (70.70) we get

u22 u˚12 “ qu˚12 u22 ´ qp1 ´ q 2 qu˚11 u21 (70.71)

which is (70.68d) with n “ 1. We proceed now by induction over n. Using the relations

u22 u˚12 “ qu˚12 u22 ´ p1 ´ q 2 qu21 u˚11 (70.72a)

u21 u˚12 “ u˚12 u21 (70.72b)

we find
“ ‰
u22 pu˚12 qn`1 “ q n pu˚12 qn u22 ´ q 2´n p1 ´ q 2n qpu˚12 qn´1 u˚11 u21 u˚12
“ q n`1 pu˚12 qn`1 u22 ´ q n p1 ´ q 2 qpu˚12 qn u21 u˚11 ´ q 2´n p1 ´ q 2n qpu˚12 qn´1 u˚11 u12 u21 .
(70.73)
In the second term we substitute u21 u11 “ qu11 u21 and in the third one we substitute u11 u˚12 “
˚ ˚ ˚

q ´1 u˚12 u˚11 . What we get is

“ ‰
u22 pu˚12 qn`1 “ q n`1 pu˚12 qn`1 u22 ´ q n`1 p1 ´ q 2 q ` q 1´n p1 ´ q 2n q pu˚12 qn u˚11 u21 (70.74a)
“ q n`1 pu˚12 qn`1 u22 ´ pq 3`n ´ q 1´n qpu˚12 qn u˚11 u21 (70.74b)
1 ´ q 2pn`1q ˚ n ˚
“ q n`1 pu˚12 qn`1 u22 ´ q 2 pu12 q u11 u21 . (70.74c)
q n`1
This is equation (70.68d).
Problèmes et choses à faire

I guess the other ones are checked similarly.

By equation (70.33c) we have

u13 u31 “ u31 u13 . (70.75)

Lemme 70.15.
The elements u13 and u31 are normal.

Proof. What we have to show is u˚13 u13 “ u13 u˚13 (definition 69.18). The relations (70.53d) and
(70.52) provide

u˚13 u13 “ u13 u˚13 ` pq 2 ´ 1qpu ˚ ˚ 2

23 u23 ` u33 u33 q ` p1 ´ q qpu
loooooooomoooooooon
˚ ˚
11 u11 ` u12 u12 q
loooooooomoooooooon (70.76a)
“1´u˚
13 u13 “1´u13 u˚
13

“ u13 u˚13 ` pq 2 ´ 1qpu13 u˚13 ´ u˚13 u13 q. (70.76b)

Thus
u˚13 u13 ´ u13 u˚13 “ pq 2 ´ 1qpu13 u˚13 ´ u˚13 u13 q (70.77)
and the combination u˚13 u13 ´ u13 u˚13 vanishes.

Let fi : SUq p3q Ñ BpHq be a representation. We write xij “ fipuij q.

Problèmes et choses à faire

Check if x1n and xn1 are normal for every n and that x1n xn1 “ xn1 x1n .
Problèmes et choses à faire
I think that looking very hard to [506] can help to justify the fact that x13 and x31 have a basis of eigenvectors. Since they are commuting, we
have a common basis of eigenvectors.
In the remaining I suppose that x13 and x31 have an unique basis of common eigenvectors.

In [564], they ask to see at [566].

70.3. REPRESENTATIONS OF SUQ p3q 2745

Proposition 70.16.
Let h P Hzt0u be a common eigenvector of x13 and x31 , namely suppose x13 h “ ⁄h and x31 h “ flh.
Then
⁄ ⁄
x13 px11 hq “ px11 hq x13 px12 hq “ px12 hq (70.78a)
q q
fl fl
x31 px11 hq “ px11 hq x31 px21 hq “ px21 hq. (70.78b)
q q
Proof. This is nothing else than the relations

qu13 u11 “ u11 u13

u12 u13 “ qu13 u12 (70.79)
u11 u31 “ qu31 u11 .

Let us suppose that fl ‰ 0 ‰ ⁄ and iterate the equations of proposition (70.16):

fl n
x31 pxn21 qh “ x h. (70.80)
q n 21
So if not vanishing, the vector xn21 h is an eigenvector for x31 with eigenvalue fl{q n . Since q † 1,
we must have k P such that xk21 h ‰ 0 and xk`1
21 h “ 0, if not we would have arbitrary large
eigenvalues for x31 , which is impossible because }x13 } § 1 by proposition 70.11.
Doing the same with x11 and the eigenvector xk21 h of x31 we find l P such that

xl11 xk12 ‰ 0 (70.81)

and
11 x12 “ 0.
xl`1 (70.82)
k

We name h1 that vector: h1 “ xl11 xk21 h. It satisfies

x11 h1 “ 0 (70.83a)
x21 h1 “ 0 (70.83b)
fl
x31 h1 “ h1 . (70.83c)
q l`k

Indeed by construction x11 h1 “ xl`1

11 x21 h “ 0 and
k

1 l k`1
x21 h1 “ x21 xl11 xk21 h “ x x h“0 (70.84)
q l 11 21
because (70.33b) implies qx21 x11 “ x11 x21 .

Proposition 70.17.
We have }x31 } “ 1 and every eigenvalue of x31 is of the form q n for some n P .

Proof. Using the same way as in the proof of proposition 70.11, we have, for every f P H ,
ÿ ÿ
}f }2 “ x x˚ki xki f, f y “ }xki f }2 . (70.85)
k k

Taking f “ h1 and i “ 1 we find

}h1 }2 “ }x11 h1 }2 ` }x21 h1 }2 ` }x31 h1 }2 “ }x31 h1 }2 . (70.86)

Thus
x13 v }x31 h1 }
}x31 } “ sup • “ 1. (70.87)
vPH }v} h1
2746 CHAPTER 70. COMPACT QUANTUM GROUPS

From proposition 70.11 we have }x31 } § 1 and thus }x31 } “ 1.

Comparing (70.86) with (70.83c) we have
ˇ ˇ
ˇ fl ˇ
}h1 } “ ˇ l`k ˇˇ }h1 },
ˇ (70.88)
q
so that |fl{q l`k | “ 1.

Let h2 be an eigenvector of x31 with eigenvalue fl2 with |fl2 | “ 1. We have

fl2
x31 x11 h2 “ x11 h2 , (70.89)
q
so that x11 h2 has to be an eigenvector of x13 with eigenvalue fl{q. Since |fl{q| ° 1 we have x11 h2 “ 0.
In the same way we find x21 h2 “ 0.
Let m P be such that xm 12 h2 ‰ 0 and x12 h2 “ 0 and set
m`1

h0 “ xm
12 h2 . (70.90)
Using the relations
u11 u12 “ qu12 u11 (70.91a)
u21 u12 “ u12 u21 (70.91b)
u31 u12 “ u12 u31 (70.91c)
we have
x11 h0 “ x21 h0 “ x12 h0 “ 0 (70.92)
and, if we set fl0 “ fl2 ,
x31 h0 “ fl0 h0 (70.93)
with |fl0 | “ 1.
In addition, if x13 h0 “ ⁄0 h0 , we have |⁄0 | “ q 2 . In order to prove this claim, we need some
more relations. Applying (70.64a) to h0 we have
x11 x˚11 h0 “ p1 ´ q 2 qh0 . (70.94)
Now we apply to h0 the equality
x11 x˚11 ` x12 x˚12 ` x13 x˚13 “ 1. (70.95)
If x13 h0 “ ⁄0 h0 we have x13 x˚13 h0 “ ⁄0 ⁄0 h0 “ |⁄0 |2 h0 . Thus
x12 x˚12 h0 “ pq 2 ´ |⁄0 |2 qh0 . (70.96)
On the other hand applying (70.64b) to h0 we have
x12 x˚12 h0 “ q 2 p1 ´ q 2 qh0 . (70.97)
Thus we have |⁄0 | “ q 2 .
What we have proved up to here is the following.

Proposition 70.18.
If there exists h P H such that x13 h “ ⁄h and x31 h “ flh with ⁄ ‰ 0 ‰ fl, then there exists h0 P H
and Â, Ï P S 1 such that }h0 } “ 1,
x11 h0 “ x12 h0 “ x21 h0 “ 0 (70.98)
and
x13 h0 “ ⁄0 h0 (70.99a)
x31 h0 “ fl0 h0 , (70.99b)
with ⁄0 “ q 2 Â and fl0 “ Ï. The eigenvalues of x31 are of the form q n .
70.3. REPRESENTATIONS OF SUQ p3q 2747

Problèmes et choses à faire

Question: do we have }x13 } “ q 2 ?

We already know that the representation space H has a basis of common eigenvectors of
x13 “ fipu13 q and x31 “ fipu31 q. The representations of SUq p3q are now divided in families that
depend on the existence of such eigenvectors with simultaneously non vanishing eigenvalues. The
proposition 70.18 describes the situation where one has a common eigenvector with non vanishing
eigenvalue of x13 and x31 . The following theorem describes the representations in that case.

Théorème 70.19.
Let fi be a representation of SUq p3q on H . Suppose that it contains a common eigenvector h0 of x13
and x31 with simultaneously non vanishing eigenvalues. Let us write x13 h0 “ ⁄0 h0 , x31 h0 “ fl0 h0
with ⁄0 ‰ 0 ‰ fl0 . Then there exist Â, Ï P S 1 such that ⁄0 “ q 2 Â and fl0 “ Ï.

Moreover the representation is described in the following way: let

|0, 0, 0y “ h0 (70.100a)
1
|N, 0, 0y “ px˚ qN |0, 0, 0y (70.100b)
AN 11
1
|N, M, 0y “ M pN `1q px˚12 qM |N, 0, 0y (70.100c)
q
1
|N, M, Ly “ N L px˚21 qL |N, M, 0y (70.100d)
q AL

´± ¯1{2
where N, M, L P and AN “ N
j“1 p1 ´ q 2j q . Then the action of SUq p3q on H is given by

x11 |N, M, Ly “ p1 ´ q 2N q1{2 |N ´ 1, M, Ly (70.101a)

2M 1{2
x12 |N, M, Ly “ q N `1
p1 ´ q q |N, M ´ 1, Ly (70.101b)
2`N `M
x13 |N, M, Ly “ q Â|N, M ´ 1, Ly (70.101c)
x21 |N, M, Ly “ q N p1 ´ q 2L q1{2 |N, M, L ´ 1y (70.101d)
x22 |N, M, Ly “ ´q L`M `1
Â̄ Ï̄|N, M, Ly (70.101e)
´ ¯1{2
´ p1 ´ q 2L qp1 ´ q 2M qp1 ´ q 2pN `1q q |N ` 1, M ´ 1, L ´ 1y (70.101f)
x23 |N, M, Ly “ q L`1 p1 ´ q 2pM `1q q1{2 Ï̄|N, M ` 1, Ly (70.101g)
´ ¯1{2
´ q M `1 p1 ´ q 2L qp1 ´ q 2pN `1q q Â|N ` 1, M, L ´ 1y (70.101h)
x31 |N, M, Ly “ q N `L Ï|N, M, Ly (70.101i)
2pL`1q 1{2
x32 |N, M, Ly “ q p1 ´ q M
q Â̄|N, M, L ` 1y (70.101j)
´ ¯
´ q L 1 ´ q 2pN `1q p1 ´ q 2M q Ï|N ` 1, M ´ 1, Ly (70.101k)
x33 |N, M, Ly “ ´q L`M `1 p1 ´ q 2pN `1q qÂÏ|N ` 1, M, Ly (70.101l)
´ ¯1{2
´ p1 ´ q 2pL`1q qp1 ´ q 2pM `1q q |N, M ` 1, L ` 1y, (70.101m)
2748 CHAPTER 70. COMPACT QUANTUM GROUPS

and

x˚11 |N, M, Ly “ p1 ´ q 2pN `1q q1{2 |N ` 1, M, Ly (70.102a)

2pM `1q 1{2
x˚12 |N, M, Ly “q N `1
p1 ´ q q |N, M ` 1, Ly (70.102b)
x˚13 |N, M, Ly “ q 2`M `N Â̄|N, M, Ly (70.102c)
2pL`1q 1{2
x˚21 |N, M, Ly N
“ q p1 ´ q q |N, M, L ` 1y (70.102d)
2
x˚22 |N, M, Ly “ ´q L`M `1
Ï|Â| |N, M, Ly (70.102e)
´ ¯1{2
´ p1 ´ q 2pL`1q qp1 ´ q 2pM `1q qp1 ´ q 2N q |N ´ 1, M ` 1, L ` 1y (70.102f)
x˚23 |N, M, Ly “ q L`1 p1 ´ q 2M q1{2 Ï|N, M ´ 1, Ly (70.102g)
´ ¯1{2
´ q M `1 p1 ´ q 2N qp1 ´ q 2pM `1q q Â̄|N ´ 1, M ` 1, Ly (70.102h)
x˚31 |N, M, Ly “ q N `L p1 ´ q 2L q1{2 Â|N, M, L ´ 1y (70.102i)
´ ¯1{2
q L p1 ´ q 2N qp1 ´ q 2pM `1q q Ï̄|N ´ 1, M ` 1, Ly (70.102j)
x˚33 |N, M, Ly “ q L`M `1 p1 ´ q 2N q1{2 Â̄ Ï̄|N ´ 1, M, Ly (70.102k)
´ ¯1{2
´ p1 ´ q 2L qp1 ´ q 2M q |N, M ´ 1, L ´ 1y. (70.102l)

The basis t|N, M, Lyu of H is orthonormal, that is

xL, M, N |A, B, Cy “ ”AN ”BM ”CL . (70.103)

Before to pass to the proof, we give some formulas that are used in the proof.

Lemme 70.20.
Under the assumptions of theorem 70.19 we have
(1) u11 pu˚12 qM h0 “ 0.

Proof. We proof (1) by induction over M . For M “ 1 we apply the relation

1
u11 u˚12 “ u˚12 u11 ` pq ´1 ´ qqpu˚22 u21 ` u32 u31 q (70.104)
q
to h0 . By construction x11 h0 “ x21 h0 “ 0. Since qu31 u˚32 “ u˚32 u31 , the vector u˚32 h0 is an
eigenvector of x31 with eigenvalue fl0 {q. Thus

u˚32 h0 “ 0 (70.105)

because |fl0 {q| ° 1. Item (1) is proved.

proof of theorem 70.19. Existence of h0 , Â and Ï are part of proposition 70.18. The action of
SUq p3q on the basis |N, M, Ly is computed using the commutation relations and the lemmas 70.13
and 70.14. That computation proves in the same time that the vectors |N, M, Ly actually form a
basis.
For sake of simplicity in the computations we introduce the vectors

h1N,M,L “ px˚21 qL px˚12 qM px˚11 qN h0 (70.106)

or
h1N,M,L “ AN AL q M pN `1q q N L|N, M, Ly. (70.107)
Let us perform the computation for (70.101a). First we compute x11 h1N,M,0 :

x11 h1N,M,0 “ x11 pX12

˚ M ˚ N
q px11 q “ q N M x11 px˚11 qN px˚12 qM h0 . (70.108)
70.4. QUANTIZED UNIVERSAL ALGEBRAS 2749

We used x˚12 x˚11 “ qx˚11 x˚12 . Using lemma 70.14 we get

´ ¯
x11 h1N,M,0 “ q M N p1 ´ q 2N qpu˚11 qN ´1 ` q 2N pu˚11 qN u11 pu˚12 qM h0 (70.109)

Using lemma 70.20 it remains

x11 h1N,M,0 “ q M N p1 ´ q 2N qpx˚11 qN ´1 px˚12 qM h0 (70.110a)

1
“ q M N p1 ´ q 2N q M pN ´1q pu˚12 qM pu˚11 qN ´1 (70.110b)
q
2N 1
M
“ q p1 ´ q qhN ´1,M,0 . (70.110c)

Using now the relation u11 u˚21 “ qu˚21 x11 we have

x11 h1N,M,L “ x11 px˚21 qL h1N,M,0 (70.111a)

“ q L px˚21 qL x11 h1N,L,0 (70.111b)
2
“q L`M
p1 ´ q N qh1N ´1,M,N . (70.111c)

Using the conversion factor (70.107) we get the desired action:

x11 |N, M, Ly “ p1 ´ q 2N q1{2 |N ´ 1, M, Ly. (70.112)

70.4 Quantized universal algebras

We follow [479].
Let g be a complex simple Lie algebra of rank n with its standard Lie bialgebra structure
(proposition 52.55). Let q P zt0u a complex number which is not a root of unity 3 .
2p– ,– q
We denote by h a Cartan subalgebra of g, by –1 , . . . , –n the simple roots and Aij “ p–ii,–ijq
the Cartan matrix 4 . We also introduce the following notations:

qi “ q p–i ,–i q{2

pa; tqk “ p1 ´ aqp1 ´ atq . . . p1 ´ atk´1 q
ˆ ˙ (70.113)
m pt; tqm
“ .
n t pt; tqn pt; tqm´n
The latter are the t-binomial coefficients also called Gauss polynomial.

Définition 70.21.
Let g be a simple complex Lie algebra with its Cartan matrix
p–i , –j q
Aij “ 2 . (70.114)
p–i , –i q
The Hopf algebra Uh g on vhw is the algebra generated by Xi , Yi , Hi (i “ 1, . . . , n) which is
complete for the h-adic topology and subject to the relations
rHi , Hj s “ 0
rHi , Xj˘ s “ ˘p–i , –j qXj˘
` ˘
` ´ sinh h2 Hi
rXi , Xj s “ ”ij ` ˘ (70.115)
sinh h2
1´Aij ˆ ˙
ÿ 1 ´ Aij
p´1qk pXi˘ qk Xj˘ pXi˘ q1´Aij ´k “ 0 if i ‰ j.
k“0
k
3. There are no n P such that q n “ 1.
4. This is a different convention from [479] in which Aij “ 2p–i , –i q{p–j , –j q.
2750 CHAPTER 70. COMPACT QUANTUM GROUPS

Proposition 70.22.
The following defined a structure of Hopf algebra on Uh g:

pHi q “ Hi b 1 ` 1 b Hi , pXi˘ q “ q ´Hi {2 b Xi˘ ` Xi˘ b q Hi {2 ,

SpHi q “ ´Hi , SpXi˘ q “ ´qi˘1 Xi˘ , (70.116)
‘pHi q “ 0, ‘pXi˘ q “ 0.

Lemme 70.23.
The algebra Uh g accepts the set of generators tHi , Ei , Fi u with
#
Ei “ Xi` q ´Hi {2 (70.117a)
Fi “ Xi´ q Hi {2 . (70.117b)

The coproduct is given by

pEi q “ Ei b 1 ` q ´Hi b Ei (70.118a)

pFi q “ 1 b Fi ` Fi b q Hi
, (70.118b)

the antipode is

SpEi q “ ´q Hi Ei (70.119a)
Spfi q “ ´Fi q ´Hi
, (70.119b)

and the counit is

‘pEi q “ 0 (70.120a)
‘pFi q “ 0. (70.120b)

Proof. In order to see how it works, we show the formula ∞ fork pEi q. Problèmes et choses à faire
One has to check the others. We consider the expansion q “
k ak x . We have
x

ÿ
pa b bqq Xb1 “ ak pa b bqpX b 1qk (70.121a)
k
ÿ
“ ak aX k b b (70.121b)
k
“ aq X b b. (70.121c)

In the same way pa b bqq 1bX “ a b bq X . Using these formulas we find

` ˘ 1 1
pEi q “ pXi q pq ´Hi {2 q “ q ´Hi {2 b Xi˘ ` Xi˘ b q Hi {2 q ´ 2 pHi b1q q ´ 2 p1bHi q (70.122a)
“ q ´Hi {2 q ´Hi {2 b Xi˘ q ´Hi {2 ` Xi˘ q ´Hi {2 b q Hi {2 q ´Hi {2 (70.122b)
“q ´Hi
b Ei ` Ei b 1. (70.122c)

If V1 and V2 are vhw-modules we define V1 b̂V2 as the limit

V1 b̂V2 “ lim V1 {hn V1 b V2 {hn V2 . (70.123)

–n

This is the completion of V1 b V2 for the h-adic topology.

Problèmes et choses à faire

I’ve to know if that limit is the one defined in section 47.2.

Théorème 70.24.
Let g be a simple complex finite dimensional Lie algebra with its standard Lie bialgebra structure.
70.4. QUANTIZED UNIVERSAL ALGEBRAS 2751

(1) There exists an algebra isomorphism

Uh g Ñ pU gqvhw (70.124)

which is the identity modulo h vhw.

(2) That isomorphism can be chosen in such a way to be the identity on h.
(3) The center of Uh g is isomorphic to Zvhw if Z is the center of U g.

For a proof, see [479], page 60.

Problèmes et choses à faire

If you understand the proof, please write me an email.

A Uh g-module V is finite dimensional if it is a -module of finite type. The corresponding

homomorphism fi : Uh g Ñ EndpV q is a finite dimensional representation.
This theorem allows to bring the representation theory of the classical case (U g) to the quantum
case. More precisely if Mh is the category of finite dimensional Uh g-modules and if Mg is the one
of finite dimensional g-modules, we can define a functor

F : Mg Ñ Mh
(70.125)
V ބ V vhw

and the action of Uh g on V vhw is the composition of the action of pU gqvhw and  where

 : Uh g Ñ pU gqvhw (70.126)

is the isomorphism of theorem 70.24.

We also consider the functor
G : Mh Ñ Mg
(70.127)
GpW q “ W {hW.
The action of U g on W {hW is given by
“ ‰
X · rvs “ Â ´1 pXq · v (70.128)

where v P W , rvs is the class modulo vhw and  is the isomorphism of theorem 70.24. Notice that
the action is well defined since

 ´1 pXq · pv ` hwq “  ´1 pXq · v ` h ´1 pXq · w. (70.129)

The last term belongs to hW .

Proposition 70.25.
The functors F and G are a bijections between the isomorphisms classes of objects of Mg and Mh .
The simple modules in Mg correspond to the indecomposable modules in Mh .

The reference [479] says that the proof is straightforward. I didn’t tried 5 .
Let P` be the set of dominant weights of g. We denote by L0 p q the U g-module of highest
weight P P` . Then we denote by Lp q such that

L0 p q “ Lp q{hLp q. (70.130)

The corresponding representation is written fi . If “ Êi is the ith fundamental weight (i.e.

Êi pHj q “ ”ij ) then we say that fiÊi is the ith fundamental representation of Uh g.

Lemme 70.26.
The module Lp q defined by L0 p q “ Lp q{hLp q is equivalently given by

Lp q “ L0 p qvhw. (70.131)
5. Let me know if it is straightforward;)
2752 CHAPTER 70. COMPACT QUANTUM GROUPS

Proof. The quotient Lp q{hLp q consist in removing all the terms of nonzero order in h. Thus

Lp q
Lp q “ vhw. (70.132)
hLp q

Proposition 70.27.
Let be a dominant weight and L0 p q be the U g-module of highest weight . We consider the
Uh g-module Lp q “ L0 p qvhw. We have the weight decomposition
à
L0 p q “ L0 p q⁄ (70.133)
⁄PP p⁄q

where P p q is the set of weights of L0 p q and

L0 p q⁄ “ tv P L0 p q tel que av “ ⁄paqv @a P hu. (70.134)

Correspondingly we have the decomposition

à
Lp q “ Lp q⁄ (70.135)
⁄PP p q

where
Lp q⁄ “ L0 p q⁄ vhw “ tv P Lp q tel que av “ ⁄paqv @a P h Ä Uh gu. (70.136)

Proposition 70.28.
The finite dimensional indecomposable Uh g-module Lp q is generated by a vector v such that
"
Xi` v “ 0 (70.137a)
Hi v “ pHi qv . (70.137b)

This vector is the highest weight vector and is the highest weight of Lp q.
In the same way, the module Lp q is generated by a vector v ˜ such that
#
Xi´ v ˜ “ 0 (70.138a)
Hi v ˜ “ ˜ pHi qv ˜ (70.138b)

where ˜ “ w0 and w0 is an element of the Weyl group of longest length. The vector v ˜ is a
lowest weight vector of Lp q.
The vectors v and v ˜ are unique up to scalar multiple.

70.4.1 Example with slp2, q

The classical representations of slp2, q are given in section 51.12 and equations (51.124). In
the case of Uh slp2, q, the representation space is given by Vm vhw and if v P Vm vhw, Z P Uh slp2, q
we define the corresponding representation

Z · v “ ÂpZq · v (70.139)

where ÂpZq U slp2, qvhw has a well defined action on v P Vm vhw. In order to determine the
representation, we have to write the isomorphism Â.
70.5. QUANTUM UNIVERSAL ENVELOPING ALGEBRA 2753

70.5 Quantum universal enveloping algebra

One speak about the representations of Uq g in the chapter 10 of [567].

Définition 70.29.
The quantum universal enveloping algebra Uq g is the complex unital algebra with generators
Xi` , Xi´ , Ki and Ki´1 (i “ 1, . . . , n) and the relations
(1) Ki Ki´1 “ Ki´1 Ki “ 1;
(2) rKi , Kj s “ 0;
Ki2 ´Ki´2
(3) rXi` , Xj´ s “ ”ij qi ´qi´1
;

(4) Ki Xj˘ “ q ˘p–i ,–j q Xj˘ Ki ;

(5)
1´Aij ˆ ˙
ÿ 1 ´ Aij
p´1q k
pXi˘ qk Xj˘ pXi˘ q1´Aij ´k “ 0 (70.140)
k“0
k qi

where qi “ q p–i ,–i q{2 .

We will sometimes write Xi for Xi` and Yi for Xi´ .

Remarque 70.30.
In the literature we find other conventions. In [458] the algebra Uq g has the relations

Ki ´ Ki´1
rEi , Fj s “ ”ij (70.141)
q ´ q ´1
and
Ki Ej Ki´1 “ q Aij Ej . (70.142)
This correspond to an other choice of q and K Ñ K 2.

Proposition 70.31.
The algebra Uq g becomes a Hopf algebra with the definitions

pKi q “ Ki b Ki pXi˘ q “ Xi˘ b Ki ` Ki´1 b Xi˘ (70.143a)

SpKi q “ Ki´1 SpXi˘ q “ ´qi˘1 Xi˘ (70.143b)
‘pKi q “ 1 ‘pXi˘ q “0 (70.143c)

where qi “ q p–i ,–i q{2 .

Proof. We have to check the relations of definition 52.26 on the generators. Let us begin by
pId b‘q “ Id:

pId b‘q Ki “ pId b‘qpKi b Ki q “ Ki b 1 “ Ki (70.144a)

pId b‘q Xi` “ pId b‘qpXi` b K ` K ´1 b Xi` q (70.144b)
“ Xi` b 1 ` K ´1 b 0 (70.144c)
“ Xi` . (70.144d)

We also have

µpId bSq Xi` “ µpXi` b SK ` K ´1 b SXi` q (70.145a)

“ Xi` K ´1 ´ qi K ´1
Xi` (70.145b)
“0 (70.145c)
“ ÷‘Xi` . (70.145d)

We used the relation ((4)) of definition 70.29.

2754 CHAPTER 70. COMPACT QUANTUM GROUPS

We denote by Uq b` the Hopf subalgebra of Uq g generated by tXi` , Ki , Ki´1 ui“1,...,n and by

Uq b´ the one generated by tX ´ , Ki , Ki´1 ui“1,...,n . The Hopf subalgebra generated by tKi , Ki´1 ui
will be denoted by Uq h.
We denote by Uq n` the subalgebra generated by tXi` ui and by Uq n´ the one generated by
tXi u. These two subalgebras are not Hopf subalgebras since the coproduct of Xi˘ involves Ki .
´

For each ‹ P we will denote by pUq n` q‹ the subspace of Uq n` generated by the monomials of
length ‹:
Xi`1 Xi`2 . . . Xi`‹ . (70.146)

70.5.1 Admissible modules

If V is a finite dimensional Uq g-module and if ⁄ P h˚ we define

V⁄ “ tv P V tel que Ki v “ q p⁄,–i q vu. (70.147)

Définition 70.32.
A finite dimensional Uq g-module is admissible if it accepts the decomposition
à
V “ V⁄ . (70.148)
⁄Ph˚

If V and W are admissible Uq g-modules we define

ÂV,W : V b W Ñ V b W
(70.149)
v b w fiÑ q p⁄,µq v b w

if v P V⁄ and w P Wµ .

Théorème 70.33.
There exists an element P Uq gb̂Uq g such that
∞
(1) reads as a sum “ 8 ‹“0 ‹ with 0 “ 1 b 1 and

‹ P pUq n` q‹ pUq hq b pUq n´ q‹ pUq hq; (70.150)

(2) for every pair of modules V and W we have

ÂV,W paq “ 1
paq ÂV,W (70.151)

where 1 “‡˝ .

Problèmes et choses à faire

Write me if you know a proof of that.

Let an admissible module à

V “ V⁄ (70.152)
⁄Ph˚

with
V⁄ “ tv tel que Ki v “ q p⁄,–i q vu. (70.153)
Let v P V⁄ , using the commutation relations we have

Ki Xj` v “ q p–i ,–j q Xj` Ki v (70.154a)

“ q p–i ,–j q q p⁄,–i q Xj` v (70.154b)
“ q p–j `⁄,–i q Xj` v. (70.154c)

Thus we have
Xj` V⁄ Ä V–j `⁄ . (70.155)
70.5. QUANTUM UNIVERSAL ENVELOPING ALGEBRA 2755

70.5.2 Example on Uq slp2, q

For notational simplicity we write

q n ´ q ´n
rns “ (70.156)
q ´ q ´1

and
rns! “ r1sr2s . . . rns. (70.157)
We follow [458] and we consider the algebra Uq slp2, q defined by the generators E, F , K,
K ´1 and the relations

KK ´1 “ K ´1 K “ 1 (70.158a)
KEK ´1 “ q 2 E (70.158b)
KF K ´1
“q ´2
F (70.158c)
K´ K ´1
rE, F s “ . (70.158d)
q ´ q ´1

The first point is to show that this family of algebra is the same as the one defined in definition 70.29
in which we pose p–, –q “ 12 because of equation (51.210). For this purpose we temporally rewrite
the equations (70.158) as

HEH ´1 “ r2 E (70.159a)
HF H ´1
“r ´2
F (70.159b)
H´ H ´1
rE, F s “ (70.159c)
r ´ r´1

and we consider the map ÂpHq “ K 2 , ÂpEq “ aX, ÂpF q “ bY . We have

ÂpHEH ´1 q “ aK 2 XK ´1 “ aKq 1{2 XK ´1 “ aqX, (70.160)

and on the other hand r2 ÂpEq “ ar2 X, so that we must have q “ r2 . As for the commutator we
have
K 2 ´ K ´2
rÂpEq, ÂpF qs “ abrX, Y s “ ab 1{4 , (70.161)
q ´ q ´1{4
so we need to fix a and b in such a way that

ab 1
“ . (70.162)
q 1{4 ´q ´1{4 r ´ r´1

The algebras (70.158) is then the same as the algebra (70.29) with g “ slp2, q.

Lemme 70.34.
There is an unique automorphism of Uq slp2, q such that ÊpEq “ F , ÊpF q “ E and ÊpKq “ K ´1 .

Proof. Unicity is automatic since Ê is defined on the generators. The point is only to see that it
extends as an automorphism. As an example we have ÊpKEK ´1 q “ K ´1 F K while q 2 ÊpEq “ q 2 F ,
but equation (70.158c) says that K ´1 F K “ q 2 F .
The other relations are checked in the same way.

Lemme 70.35.
Let m • 0 and n P . We have the following relations:

E m K n “ q ´2mn K n E m (70.163a)
2mn
m
F K “q n
K F , n m
(70.163b)
2756 CHAPTER 70. COMPACT QUANTUM GROUPS

and
q ´pm´1q K ´ q m´1 K ´1
rE, F m s “ rmsF m´1 (70.164a)
q ´ q ´1
q m´1 K ´q ´pm´1q K ´1 m´1
“ rms F (70.164b)
q ´ q ´1
and
q ´pm´1q K ´ q m´1 K ´1 m´1
rE m , F s “ rms E (70.165a)
q ´ q ´1
q m´1 K ´ q ´pm´1q K ´1
“ rmsE m´1 . (70.165b)
q ´ q ´1
Proof. We check the relations (70.163a) and (70.164a). The other can be deduced applying Ê.
For the first one, we use the commutator (70.158b) under the form EK “ q ´2 KE. We have
E m K “ q ´2m KE m .
With m “ 1 the relation (70.164a) reduces to the definition of Uq slp2, q. We proceed by
induction on m: assuming that the result is true with m ´ 1 we have

rE, F m s “ rE, F m´1 sF ` F m´1 rE, F s (70.166)

in which we substitute
q ´pm´2q K ´ q m´2 K ´1
rE, F m´1 sF “ rm ´ 1sF m´2 F. (70.167)
q ´ q ´1
Grouping the terms we find
F m´1 ´ ¯
rE, F m s “ rm ´ 1spq ´m
K ´ q m ´1
K q ` K ` K ´1
(70.168)
q ´ q ´1
The coefficient with K in the parenthesis is
q m´1 ´ q ´m`1 ´m
rm ´ 1sq ´m ` 1 “ q ` 1. (70.169a)
q ´ q ´1
Taking the common denominator we find
q m ´ q ´m
q ´m`1 “ q ´m`1 rms. (70.170)
q ´ q ´1
Putting all together we find the result.

Corollaire 70.36.
If v is an eigenvector of K with eigenvalue –, and if vn “ E n v is non vanishing, then it is an
eigenvector with eigenvalue q 2n –. In particular the vectors vn are distinct.

Proof. This is a computations using the relation (70.163a):

KE n v “ q 2n E n Kv “ q 2n –E n v. (70.171)

Since g “ slp2, q, the Cartan algebra is one dimensional and the roots are just numbers. Let
V be a Uq slp2, q-module. We consider, for ⁄ ‰ 0,

V⁄ “ tv P V tel que Kv “ ⁄vu. (70.172)

We say that ⁄ is a weight is V⁄ ‰ t0u.

Let V be a Uq slp2, q-module and ⁄ P . One say that v ‰ 0 is a height weight vector of
weight ⁄ if Ev “ 0 and Kv “ ⁄v. The module V is said to be a highest weight if it is generated
by a highest weight vector.
70.5. QUANTUM UNIVERSAL ENVELOPING ALGEBRA 2757

Lemme 70.37.
We have EV⁄ Ä Vq2 ⁄ and F V⁄ Ä Vq´2 ⁄ .

Proof. If v P V⁄ the relations (70.158b) and (70.158c) imply

KpEvq “ q 2 EKv “ q 2 ⁄pEvq (70.173)

and
KpF vq “ q ´2 F Kv “ q ´2 ⁄pF vq. (70.174)

Proposition 70.38.
Every finite dimensional Uq slp2, q-module has a highest weight vector.

Proof. Since is algebraically closed, the operator K has an eigenvector. Let Kw “ –v with
– ‰ 0. If Ew “ 0, this is a highest weight vector. If not we consider wn “ E n w (n P ). By
corollary 70.36, the vectors wn are eigenvectors of K with distinct eigenvalues; thus there exists a
n such that wn ‰ 0 and wn`1 “ 0.

Lemme 70.39.
If V is a finite dimensional Uq slp2, q-module, the elements E and F are nilpotent as endomor-
phisms of V .

Proof. Let U be an unitary matrix such that U ˚ EU is upper diagonal (proposition 46.4). The
eigenvalue of E and U ˚ EU are the same and are on the diagonal of U ˚ EU . If we prove that the
eigenvalue of E are zero, then U ˚ EU will be nilpotent and thus E itself will be nilpotent.
Let Ev “ ⁄v. Since EKv “ ⁄q ´2 Kv, the vectors K n v are eigenvectors of E with distinct
eigenvalues ⁄q ´2n . This is impossible in a finite dimensional space, so ⁄ “ 0.

Proposition 70.40.
Let V be a Uq slp2, q-module and v a highest weight vector of weight ⁄. We consider the sequence
v0 “ v
1 p
vp “ F v. (70.175)
rps!
Then each time vp is nonzero we have

Kvp “ ⁄q ´2p vp (70.176a)

K ´1 vp “ ⁄´1 q 2p vp (70.176b)
q ´pp´1q ⁄ ´ q p´1 ⁄´1
Evp “ vp´1 (70.176c)
q ´ q ´1
F vp´1 “ rpsvp . (70.176d)

Proof. For the first one, using the commutation relations,

1
Kvp “ KF p v (70.177a)
rps!
1 2p p
“ q F Kv (70.177b)
rps!
1
“ ⁄q 2p F p v (70.177c)
rps!
“ ⁄q 2p vp . (70.177d)

The action of K ´1 is immediately deduced from that one applying K on both sides. For the last
one,
1 1
F vp´1 “ F pv “ rps!vp “ rpsvp . (70.178)
rp ´ 1s! rp ´ 1s!
2758 CHAPTER 70. COMPACT QUANTUM GROUPS

And for the second one we write EF p v “ rE, F p sv ` lo on in which we substitute (70.164b). We
Fopmo
Ev
“0
find
q p´1 K ´ q ´pp´1q K ´1
Evp “ vp´1 . (70.179)
q ´ q ´1
We already know the action of K and K ´1 on vp´1 .

Corollaire 70.41.
The vectors vp are eigenvectors of K with distinct eigenvalues.

Proof. This is equation (70.176a) and the fact that q is not a root of unity.

The structure of the simple Uq slp2, q-modules is described by the theorems 70.42 and 70.43.

Théorème 70.42.
Let V be a finite dimensional Uq slp2, q-module generated by a highest weight vector v of weight
⁄. Then
(1) The weight ⁄ reads ⁄ “ ‘q n where ‘ “ ˘1 and n “ dimpV q ´ 1.
1
(2) Set v0 “ v and vp “ p
rps! F v. We have vp “ 0 with p ° n and the set

tv0 , v1 , . . . , vn u (70.180)

is a basis of V .
(3) The action of K on V is diagonalisable and has the m ` 1 distinct eigenvalues

t‘q n , ‘q n´2 , ¨ ¨ ¨ , ‘q ´n`2 , ‘q ´n u. (70.181)

(4) Any highest weight vector in V is a multiple of v.

Théorème 70.43.
A Uq slp2, q-module is simple if and only if it is generated by an highest weight. Moreover if two
modules are generated by a highest weight vector of same weight, they are isomorphic as Uq slp2, q-
modules.

Proof of theorem 70.42. We know that the vectors vp are eigenvector of K with distinct eigenvalues.
Thus there exits n P such that vn “ 0 and vn`1 “ 0. For that value of n we have vm “ 0 @m ° n
and vm ‰ 0 when m § n. We also have
q ´n ⁄ ´ q n ⁄´1
0 “ Evn`1 “ vn , (70.182)
q ´ q ´1
so that q ´n ⁄ ´ q n ⁄´1 “ 0. This implies ⁄ “ ˘q n .
Let us now prove that tv0 , . . . , vn u is a basis of V . This will show that dimpV q “ n ` 1. First
we know that the vectors vi are eigenvectors of K for distinct eigenvalues, so that the set tvi u is
free. By hypothesis, the vector space V is generated by v0 . From the relations (70.176), we see
that the action of Uq slp2, q on the vectors vi will only generate linear combinations of the vectors
vi . Thus the set of vi ’s is generating. The points (1) and (2) are proved.
For item (3), the operator K is diagonalisable because the vectors vp form a basis of eigenvectors
of K. The eigenvalues are given by the relation Kvp “ ⁄q ´2p vp . Since ⁄ “ ‘q n we have the
eigenvalues
t‘q n , ‘q n´2 , . . . , ‘q ´n u (70.183)
corresponding to p “ 0, . . . , n.
In order to prove point (4) let v 1 be an other highest weight vector. By definition Ev 1 “ 0
and there exists an ⁄1 P such that Kv 1 “ ⁄1 v 1 . Since v 1 is eigenvector of V , this is up to scalar
multiple one of the vectors vi . From the constraint Evi “ 0 we see that i “ 0 so that v 1 is multiple
of v0 “ v.
70.6. QUANTIZED FUNCTION ALGEBRA 2759

Proof of theorem 70.43. First we suppose that V is generated by an highest weight v and we prove
that it is simple. LetV 1 be a submodule of V . By proposition 70.38 V 1 has an highest weight
vector v 1 . The vector v 1 is highest weight for V also, thus v 1 is a multiple of v. Since v P V 1 we
have V Ä V 1 and consequently V “ V 1 .
Let V be a simple module, v a highest weight vector in V and V 1 the subspace of V generated
by v. The space V 1 is a submodule of V while V is simple, so V 1 “ V and v generated V .
Finally let V be generated by v of weight ⁄ and V 1 be generated by v 1 of same weight ⁄. Since
1
⁄ “ ‘q n “ ‘1 q n (70.184)

we have ‘ “ ‘1 and 6 n “ n1 , so the modules V and V 1 have same dimension n ` 1. The set
tvp “ F p v{rps!u is a basis of V while the set tvp1 “ F p v 1 {rps!u is a basis of V 1 . One checks that the
map  : V Ñ V 1 , Âpvi q “ vi1 is an isomorphism of Uq slp2, q-modules.

The conclusion is that the Uq slp2, q-modules are classified up to isomorphisms by their di-
mension n ` 1 and ‘ “ ˘1. We denote by V‘,n the module of dimension n ` 1 with highest weight
⁄ “ ‘q n . The corresponding representation fl‘,n : Uq slp2, q Ñ EndpV‘,n q is given by

Kvp “ ‘q n´2p vp
Evp “ ‘rn ´ p ` 1svp´1 (70.185)
F vp´1 “ rpsvp

with p “ 0, . . . , n.

70.6 Quantized function algebra

Définition 70.44.
Let G be the connected simply connected Lie group with Lie algebra g. The quantized algebra of
regular functions on the group G is the Hopf ˚-subalgebra of matrix elements of the unitarizable
finite dimensional Uq g-modules. We denote it by rGsq

The multiplication law in rGsq is given by formula (52.94a):

1
cl1 ,v1 cl,v pxq “ pl1 b lqfl 1b pxqpv 1 b vq “ pl1 b lqp xqpv 1 b vq, (70.186)

so
1 1b
cl1 ,v1 cl,v “ cl1 bl,v 1 bv (70.187)
where 1b stands
∞ for the representation of Uq g b Uq g on Lp q. We can be more explicit
1 q b Lp

if we write x “ i ai b bi , we have
1
ÿ 1
ÿ ` ˘ ` ˘
cl,v cl1 ,v1 pxq “ cl,v pai qcl1 ,v1 pbi q “ l fl pai qv l1 fl 1 pbi qv 1 (70.188)
i i
1
where fl stands for the representation on Lp q. At the end the product cl1 ,v1 cl,v is still an element
in rGsq and so a linear form on Uq g.

Remarque 70.45.
In the setting of Hopf algebra we consider the notion of involution of the definition 52.38 in which
we do not require a˚˚ “ a.

Proposition 70.46.
If q P 0 the map Ê : Uq g Ñ Uq g defined on the generators by

ÊpXi˘ q “ Xi¯ (70.189a)

ÊpKi q “ Ki . (70.189b)
6. Once again we use the fact that q is not a root of unity.
2760 CHAPTER 70. COMPACT QUANTUM GROUPS

One also speaks about that structure in [568].

Since Uq g is a Hopf ˚-algebra, the dual also becomes a Hopf ˚-algebra by proposition 52.39.
The algebra pUq gq˚ contains in particular the matrix elements of Uq g-modules, so the space rGsq
is a Hopf ˚-algebra.
Let V “ Lp q be the simple admissible Uq g-module of highest weight P P` where P` is the
set of dominant weights in h˚ . Since it is admissible we have the decomposition
à
Lp q “ Lp q⁄ (70.190)
⁄Ph˚

with
Lp q⁄ “ tv P Lp q tel que Ki v “ q p⁄,–i q vu. (70.191)

By general theory (see subsection 52.6.6), the dual vector space Lp q˚ is also a Uq g-module. We
chose on Lp q˚ the left Uq g-module structure pa · lq “ Lpaql in the sense of definition (52.119),
that means
` ˘
pa · lqpvq “ l S ´1 paqv (70.192)

for any a P Uq g, l P Lp q˚ and v P Lp q.

Lemme 70.47.
The module Lp q˚ accepts the decomposition
à
Lp q˚ “ Lp q˚⁄ (70.193)
⁄Ph˚

` ˘˚
where Lp q˚⁄ “ Lp q´⁄ .

` As vector
Proof. ˘˚ space the decomposition (70.193) is nothing else that pA ‘ Bq˚ “ A˚ ‘ B ˚ . Let
– P Lp q⁄ . By definition (52.119) and using the antipode (70.143b) if v P Lp q⁄ we have
` ˘ ` ˘
pKi –qv “ – S ´1 pKi qv “ –pKi´1 vq “ – q ´p⁄,–i q v “ q ´p⁄,–i q –pvq. (70.194)

Thus
Ki – “ q ´p⁄,–i q –. (70.195)

This proves that

` ˘˚ ` ˘
Lp q⁄ “ Lp q˚ ´⁄ . (70.196)

From a notational point of view, what we write Lp q˚⁄ is

` ˘ ` ˘˚
Lp q˚⁄ “ Lp q˚ ⁄ “ Lp q´⁄ (70.197)

We can consider the regular left and right representations of Uq g on rGsq , since the latter is
a part of the dual of Uq g. The space rGsq becomes a Uq g b Uq g-module with the representation
L b R described around equation (52.125).

Proposition 70.48.
The map
à
Â: rGsq Ñ Lp q˚ b Lp q
PP` (70.198)
Lp q
cl,v ބ l b v

is an isomorphism of Uq g b Uq g-modules. Here P` is the set of dominant weights of g.

70.6. QUANTIZED FUNCTION ALGEBRA 2761

Proof. The fact that  is bijective is contained in the definition of rGsq . The point is to check
that this is a morphism. Let a b b P Uq g b Uq g. The same computation as in (52.126) shows that
` ˘ ` ˘ ` ˘ ` ˘
pa b bq · cl,v pxq “ LpaqRpbqcl,v pxq “ cl,v S ´1 paqxb “ l S ´1 paqxbv “ cLpaql,bv pxq, (70.199)

so we have
Âpa b bq · cl,v “ Lpaql b bv. (70.200)
On the other hand we have

pa b bqÂcl,v “ pa b bql b v “ Lpaql b bv. (70.201)

The map  is then an homomorphism.

The fact that the sum in (70.198) only runs over the dominant weight comes from subsec-
tion 51.23.3.

Proposition 70.49.
Let w0 be the element of the Weyl group which sends positive roots to negative ones 7 . Then we
have
1 1
c´ ,⁄ c´w0 1 , 1 “ q ´2x ,⁄y c´w0 1 , 1 c´ ,⁄ . (70.202)

Problèmes et choses à faire

Formula (70.202) is not the one given in [479], equation (2.1.4) page 98. I don’t know where I got wrong in the computation of the proof.

1 1
Proof. Consider the elements cl,v and cl1 ,v1 in c´ ,⁄ and c´w0 1, 1 with

l P Lp q˚´ , v P Lp q⁄
1 1 ˚ (70.203)
l P Lp q´w0 1 , v 1 P Lp 1 q 1 .
` ˘˚
First notice that l1 pv 1 q “ 0 because l1 P Lp 1 qw0 1 while v 1 P Lp 1 q 1 and w0 1 ‰ 1 . In the
same way fl 1 pXi` qv 1 “ 0 since v 1 is a highest weight vector in Lp 1 q.
1
The product cl,v cl1 ,v1 is defined by the equation (52.94a). If we apply it to Ki we find
1 1
pcl,v cl1 ,v1 qpKi q “ pcl,v b cl1 ,v1 qp Ki q
` ˘ ` ˘
“ l fl pKi qv l1 fl 1 pKi qv 1 (70.204)
“0
1
because fl 1 pKi qv 1 is a multiple of v 1 . The same shows that pcl1 ,v1 cl,v qpKi q “ 0.
We still have to check the equality on Xi˘ . On the left hand side of (70.202) we get
1 1
pcl,v cl1 ,v1 qpXi˘ q “ pcl,v b cl1 ,v1 qpXi˘ b Ki ` Ki´1 b Xi˘ q (70.205a)
` ˘ 1` ˘ ` ˘` ˘
“ l fl pXi˘ qv lloooooomoooooon
fl 1 pKi qv 1 `l fl pKi´1 qv l fl 1 pXi q˘ v 1 (70.205b)
“0
` ˘
“ q ´x ,⁄y
lpvql1 fl 1 pXi˘ qv 1 . (70.205c)

On the right hand side we have

1 ` ˘` ˘ ` ˘ ` ˘
pcl1 ,v1 cl,v qpXi˘ q “ l1 fl 1 pXi˘ qv 1 l flKi v ` lloooooooomoooooooon
1
fl 1 pKi´1 qv 1 l fl pXi˘ qv (70.206a)
“0
` ˘
“ qx ,⁄y 1
l fl 1 pXi˘ qv 1 lpvq.j (70.206b)

7. See theorem 51.186