arXiv:1711.00082v1 [math.

FA] 31 Oct 2017

Radial Toeplitz operators on the weighted Bergman spaces

of Cartan domains

Matthew Dawson and Raul Quiroga-Barranco

Abstract. Let D be an irreducible bounded symmetric domain with biholo-

morphism group G with maximal compact subgroup K. For the Toeplitz
operators with K-invariant symbols we provide explicit simultaneous diago-
nalization formulas on every weighted Bergman space. The expressions are
given in the general case, but are also worked out explicitly for every irre-
ducible bounded symmetric domain including the exceptional ones.

1. Introduction
Irreducible bounded symmetric domains and the weighted Bergman spaces that
they support constitute a fundamental object in functional analysis. These allow us
to consider Toeplitz operators and irreducible unitary representations of Lie groups.
In the last two decades, it has been observed that there is a natural and far reaching
relationship between these operators and representations. This work continues the
study of Toeplitz operators derived from such interaction.
For D a bounded symmetric domain with biholomorphism group G, it was
proved in [1] that there are plenty of closed subgroups H in G so that the Toeplitz
operators with H-invariant symbols pairwise commute. This yields commutative
C -algebras generated by Toeplitz operators. Ultimately, this is a consequence of
multiplicity-free restriction results. Among the latter, the very first one is found in
[15] for the maximal compact subgroup K in G that fixes a given point.
We consider the multiplicity-free decomposition for the restriction to K of the
unitary representations of G on the Bergman spaces, as found in [15] or [18].
This is used to achieve our main goal: an explicit simultaneous diagonalization of
Toeplitz operators with K-invariant symbols. A general but also very explicit for-
mula is obtained in Theorem 4.12, where every K-invariant symbol is really seen as
a complex-valued function defined on [0, 1)r , where r is the rank of D. Correspond-
ingly, our formulas are given by integrals computed in this set. Our computations
make use of the spherical and conical polynomials for the representation of K on
the space of polynomials on the domain D. Hence, Theorem 4.11 provides the si-
multaneous diagonalization in terms of these polynomials. Since there are explicit
expressions for the conical polynomials (see [8]) this should prove to be useful in
further developments.

2010 Mathematics Subject Classification. Primary 47B35 22D10 Secondary 32M15 22E46.
Research supported by SNI and Conacyt.
1
2 MATTHEW DAWSON AND RAUL QUIROGA-BARRANCO

On the other hand, we rely strongly on the Jordan structures associated to

the irreducible bounded symmetric domains. For example, the preliminary com-
putation given by Theorem 4.9 reduces the diagonalization of a Toeplitz operator
with K-invariant symbol to integrals over the cone of the largest tube type domain
contained in D. Most importantly, the Jordan structure allows us to consider prim-
itives (a sort of minimal tripotent) thus making possible to obtain the formula in
Theorem 4.12 that involves integrals over [0, 1)r . This integration is in fact given
by taking coordinates with respect to a maximal collection of primitives (in the
Jordan setup) or a basis of a maximal flat (in the Lie setup). In the last section
we exhibit examples of these sets for every irreducible bounded symmetric domain,
including the exceptional ones, which leads us to completely explicit formulas that
provide the diagonalization of Toeplitz operators with K-invariant symbols.
The properties of Jordan pairs and analysis on irreducible bounded symmetric
domains that we require are reviewed in Sections 2 and 3, respectively. We obtain
our main results for irreducible bounded symmetric domains in the last two sections:
Section 4 with general expressions for an arbitrary domain and Section 5 with
explicit formulas for every irreducible case.

2. Jordan structures associated to bounded symmetric domains

In this section we collect some standard facts on bounded symmetric domains.
The standard references are [2, 3, 5, 6, 12, 18, 19] where the reader will find
further details and proofs.
Let D be an irreducible Hermitian symmetric space of non-compact type, and
denote by G the connected component of the automorphism group of D and by K
the subgroup of G fixing some point. Consider the corresponding Cartan decom-
position given by
g = kp
where g and k are the Lie algebras of G and K, respectively. Choose h a Cartan
subalgebra of k, which is thus a Cartan subalgebra of g. For the complexified Lie
algebras, we recall that a root of gC with respect to hC is called compact if the
corresponding root space g is contained in kC . Let us fix an order and let us denote
by + the space of all positive noncompact roots. We consider the complex space
X
p+ = g .
+

Then, it is well known that D admits a canonical embedding as a circled bounded

domain in p+ such that K is precisely the isotropy subgroup of the origin: this
is the Harish-Chandra realization of D. The complex space p+ carries a natural
Hermitian inner product obtained from the Killing form of gC and the conjugation
with respect to the real form k ip.
Every element X p+ can be considered as a holomorphic vector field over D
whose one-parameter group of transformations is t 7 exp(tX) acting on D, and so
X can be seen as a holomorphic map D p+ . Then, the vector field X can be
written as
X(z) = y Qy (z, z),
for every z D, where y = X(0) and Qy (z, w) is a complex bilinear form of (z, w).
Furthermore, for every z, w the assignment y 7 Qy (z, w) is complex anti-linear.
 RADIAL TOEPLITZ OPERATORS ON CARTAN DOMAINS 3

This yields a triple product

p+ p+ p+ p+
{zyw} = Qy (z, w),
that defines a Jordan pair which also carries a natural involutive anti-linear map
z 7 z . From now on, we will consider p+ endowed with this algebraic structure.
For this setup, an element z p+ is called tripotent if and only if {zzz} = z,
and two tripotent elements z, w are called orthogonal when {zwy} = 0 for all
y p+ . A tripotent element is called primitive if and only if it is not the sum
of two orthogonal tripotents. Let e1 , . . . , er be a maximal collection of mutually
orthogonal primitive tripotents. It follows from [12] that r is precisely the rank of
D. It will be useful to consider the element
e = e1 + + er ,
that is a tripotent element in p+ as well.
For every tripotent z there is a so-called Peirce decomposition
p+ = V1 (z) V 12 (z) V0 (z),
where
V (z) = {w p+ | {zzw} = w}.
Then, the space V1 (z) is a Jordan algebra with the product x y = {xzy} and for
this structure z is the identity.
From now on, we will denote
Vj = V1 (e1 + + ej )
for j = 1, . . . , r, which is thus a Jordan algebra with identity e1 + + ej . We
observe that
V1 Vr .
It follows from [2] that Dj = Vj D is a bounded symmetric domain of tube type
whose associated symmetric cone is given by
j = {x2 | x Vj , x = x}.
Let us denote by Lj the identity component of the isotropy subgroup of automor-
phisms of j that fix e1 + + ej . By the results from [2, 11] there is a unique
polynomial j (z) in Vj that is Lj -invariant and that satisfies
(2.1) j (t1 e1 + + tj ej ) = t1 . . . tj .
We will consider the polynomials 1 (z), . . . , r (z) as defined on p+ by orthogonal
projection.
For the case j = r we drop the index, and so we have V = Vr , = r ,
(z) = r (z). We will also denote
DT = V D,
which is a bounded symmetric domain of tube type. We will call DT the tube type
domain associated to D. In fact, DT is the largest tube type domain obtained as
the intersection of D with a complex subspace of p+ .
On the other hand, there is a joint Peirce decomposition associated to e1 , . . . , er
given by M
p+ = Vjk ,
0j,kr
4 MATTHEW DAWSON AND RAUL QUIROGA-BARRANCO

where we define
Vjj = V1 (ej ), for j = 1, . . . , r,
Vjk = V 21 (ej ) V 12 (ek ), for j, k = 1, . . . , r,
\
Vj0 = V0j = V 21 (ej ) V0 (ek ), for j = 1, . . . , r,
k6=j
V00 = V0 (e1 ) V0 (er ).
We note that since e1 , . . . , er is maximal, it turns out that V00 = 0. It is also well
known that there exists a pair of integers a > 0 and b 0 such that
dim Vjj = 1, for j = 1, . . . , r,
dim Vjk = a, for j, k = 1, . . . , r,
dim Vj0 = b, for j = 1, . . . , r.
The numbers a, b are the characteristic multiplicities of D. We observe that D is
of tube type if and only if b = 0. If we denote by n and nT the dimensions of D
and DT , respectively, then we have
r(r 1) r(r 1)
n=r+ a + rb, nT = r + a.
2 2
The genus of the domain D is defined by
n + nT
p= = 2 + (r 1)a + b.
r
Finally, it is also well known that there is a set of strongly orthogonal roots
1 , . . . , r in the sense of Harish-Chandra such that, for every j = 1, . . . , r, the
tripotent ej belongs to the root space gj . Without loss of generality, we can
assume that 1 > > r > 0.

3. Analysis on bounded symmetric domains

With the setup provided by Section 2 we have the following integral formula
(see [2, 6, 18])
Z
(3.1) f (z) dz
p+
Z Z  X r  Y r
Y
=c f g tj ej dg |t2j t2k |a t2b+1
j dt1 . . . dtr
[0,)r K j=1 1j<kr j=1

where dz is the Lebesgue measure on p+ corresponding to its Hermitian structure,

dg is the probability Haar measure of K and c is a positive constant whose value
will not be used.
We also have the following integral formula.
Z
(3.2) f (x) dx

Z Z  X r  Y
= c f s xj ej ds |x2j x2k |a dx1 . . . dxr ,
[0,)r L j=1 1j<kr

where dx is the Lebesgue measure on , dg is the Haar measure of L and c is a

constant whose value we will not need.
 RADIAL TOEPLITZ OPERATORS ON CARTAN DOMAINS 5

On the other hand, for the Lebesgue measure on p+ , we have (see [10])
Yr
(p nr (j 1) a2 )
vol(D) = n .
j=1
(p (j 1) a2 )

Hence, we consider the normalized Lebesgue measure

r
1 1 Y (p (j 1) a2 )
dv(z) = dz = n dz.
vol(D) j=1 (p nr (j 1) a2 )

The (weigthless) Bergman space is the closed subspace A2 (D) of L2 (D, v) that
consists of holomorphic functions. The reproducing kernel of this space can be
described as follows.
Let us define the Bergman endomorphisms
B(x, y) : p+ p+
B(x, y)z = z 2{xybz} + {x{yzy}x},
for any x, y p+ . Then, the reproducing KD of A2 (D) is given by (see [12])
KD : D D C
KD (z, w) = det(B(z, w))1 .
Furthermore, as noted in [2] there is a real polynomial h(z, w) on p+ p+ ,
holomorphic in z and anti-holomorphic in w, such that
h(z, w)p = det(B(z, w)).
It can also be described as the unique K-invariant polynomial such that
r
Y
2
(3.3) h(x, x) = (e x ) = (1 x2j )
j=1

for every x = x1 e1 + + xr er (xj R) and with the square x2 computed in the

Jordan algebra V . Note that we have used equation (2.1) with j = r.
Hence, the Bergman kernel can also be computed as
KD (z, w) = h(z, w)p ,
for every z, w D.
The following weighted volume computation can also be obtained (see [18])
Z Yr
( nr (j 1) a2 )
h(z, z)p dz = n ,
D j=1
( (j 1) a2 )

which holds for every > p 1. Hence, for every such we consider the weighted
measure
r
1 Y ( (j 1) 2a )
dv (z) = n h(z, z)p dz,
j=1 ( nr (j 1) a2 )
which thus satisfies v (D) = 1.
For every > p 1, the weighted Bergman space A2 (D) corresponding to is
given by
A2 (D) = {f L2 (D, v ) | f is holomorphic}.
6 MATTHEW DAWSON AND RAUL QUIROGA-BARRANCO

This is a closed subspace of L2 (D, v ) and a reproducing kernel space with Bergman
kernel given by
KD, : D D C
KD, (z, w) = h(z, w) .
In particular, the corresponding Bergman projection BD, : L2 (D, v ) A2 (D)
satisfies
Z
BD, (f )(z) = f (w)h(z, w) dv (w)
D
r Z
1 Y ( (j 1) a2 ) f (w)h(w, w)p dw
= n n a .
j=1 ( r (j 1) 2 ) D h(z, w)

We note that A2p (D) = A2 (D) is the Bergman space corresponding to the
(weightless) Lebesgue measure.
For any function L (D), the Toeplitz operator on A2 (D) with bounded
symbol is defined by
T : A2 (D) A2 (D)
T (f ) = BD, (f ).
On the other hand, for every > p 1 there is a unitary representation of the
e of G
universal covering group G
e A2 (D) A2 (D)
: G

( (g)f )(z) = J(g 1 , z) p f (g 1 z),
where J(g, z) denotes the complex Jacobian of the transformation g at the point z.
These yield the holomorphic discrete series of G.
Since K is the subgroup of linear unitary transformations on p+ that belong to
G we have J(g, z) 1 for all g K. In particular, for every > p 1 the above
yields a unitary representation of K itself given by
|K : K A2 (D) A2 (D)
( (g)f )(z) = f (g 1 z).

4. Radial operators and the isotypic decomposition of |K

Let : H U(H) be a unitary representation of a Lie group H. We recall
that a bounded operator T : H H is called -intertwining if we have
T (h) = (h) T
for every h H. The algebra of -intertwining bounded operators is denoted by
EndH (H). This algebra is a fundamental invariant to understand the decomposition
of H into irreducible subspaces. In particular, we have the following well known
result where we have considered the case of compact groups.
Proposition 4.1. The unitary representation of the group H on the Hilbert
space H is multiplicity-free if and only if EndH (H) is commutative. For H com-
pact, this occurs exactly when H can be decomposed as a direct sum of inequivalent
irreducible H-invariant closed subspaces.
 RADIAL TOEPLITZ OPERATORS ON CARTAN DOMAINS 7

By focusing our attention to the subgroup K of G and the holomorphic discrete

series we obtain a special type of operators.
Definition 4.2. A bounded operator T : A2 (D) A2 (D) is called radial if
it is |K -intertwining, i.e. if T satisfies
T (g) = (g) T
for every g K. The algebra of bounded radial operators on A2 (D) is denoted by
EndK (A2 (D)).
It is well known that the unitary representations are irreducible highest
weight representations of the group G. e The fact that K is a maximal compact
subgroup of G thus implies that the restrictions |K are multiplicity-free: this
follows, for example, from the results in [9] and [13].
As a consequence of the previous remarks we obtain the following.
Proposition 4.3. For every > p 1, the algebra EndK (A2 (D)) is commu-
tative.
We now consider the corresponding invariance property for symbols and their
associated Toeplitz operators.
Definition 4.4. A bounded symbol L (D) is called radial if it is K-
invariant. In other words, if for every g K we have
(gz) = (z)
for almost every z D.
We recall the following well known result. It follows from [5] by using the fact
that the real span of e1 , . . . , er is a maximal Abelian subspace of p+ .
Proposition 4.5. For every z D there exist g K and t [0, 1)r such that
X r
z=g tj e j .
j=1

In particular, a bounded symbol L (D) is radial if and only if we have

Xr 
(z) = tj e j ,
j=1

when the above relation holds between z D and t [0, 1)r .

It was observed in [1] the equivalence between the invariance of a symbol and of
its associated Toeplitz operators. In particular, we have the following result which
is a particular case of Corollary 3.3 and Theorem 5.2 from [1].
Proposition 4.6 (DQO [1]). For a given bounded symbol L (D) and a
fixed > p 1 the following conditions are equivalent
(1) The symbol is radial.
(2) The Toeplitz operator T on A2 (D) is radial.
Hence, if we denote by AR the space of bounded radial symbols and by T () (AR )
the C -algebra generated by the Toeplitz operators on A2 (D) with bounded radial
symbols, then we have
T () (AR ) EndK (A2 (D)).
In particular, for every > p 1 the C -algebra T () (AR ) is commutative.
8 MATTHEW DAWSON AND RAUL QUIROGA-BARRANCO

We now describe the multiplicity-free decomposition of A2 (D) with respect to

the representation |K .
Recall that the space of polynomials P(p+ ) is a dense |K -invariant subspace
of A2 (D) for every > p 1. Hence, it is enough to describe the isotypic decom-
position of the representation of K on P(p+ ) (see [15] and [18]).


Let us denote by N r the space of r-tuples of integers (1 , . . . , r ) satisfying
the condition 1 r 0. Then, the highest weights of the irreducible
K-submodules contained in P(p+ ) are precisely those of the form
(4.1) 1 1 r r
r


for some N . Furthermore, for each N r there is exactly one such K-

submodule that we will denote by P (p+ ). In particular, we have a K-invariant
algebraic direct sum M
P(p+ ) = P (p+ ),


N r
that induces the multiplicity-free isotypic decomposition for the representation |K
given by M
A2 (D) = P (p+ ).


N r
In particular, for every weight of the form (4.1) the corresponding space of highest
vectors in P(p+ ) is 1-dimensional.
The following is an immediate consequence of the previous discussion.
Proposition 4.7. For every radial operator T : A2 (D) A2 (D) and for


every N r let us denote by c (T ) the complex number such that
T |P (p+ ) = c (T )IP (p+ ) .
Then, the map defined by


End |K (A2 (D)) ( N r )
T 7 (c (T ))
r,
N

is an isomorphism of C -algebras. Furthermore, for every radial operator T we
have
hT , i
c (T ) = ,
h, i


for every non-zero P (p+ ) and for every N r .
On the other hand (see [8, 2, 18]), it is well known that the polynomial j (z)
on p+ is a highest weight vector for |K with corresponding weight 1 j ,
for every j = 1, . . . , r. In particular, the space of highest weight vectors is the free
polynomial algebra C[1 (z), . . . , r (z)]. As a consequence of this, a highest weight
vector corresponding to (4.1) is given by
(z) = 1 (z)1 2 2 (z)2 3 . . . r (z)r ,


for every N r . These are called the conical polynomials associated to |K .
We have the following naturally defined polynomial
Z
(z) = (sz) ds,
L
 RADIAL TOEPLITZ OPERATORS ON CARTAN DOMAINS 9

for every N r , which belongs to P (p+ ) and that is (L-)spherical for the repre-
sentation |K restricted to P (p+ ).
The next result is proved, for example, in [18] (see Theorem 2.8.10 therein).
Lemma 4.8. With the above notation we have
Z
1
| (kz)|2 dk = (z 2 ),
K d


for every N r and z V , where d = dim P (p+ ) and for z 2 computed using
the Jordan algebra structure of V described in Section 2.
The previous results allow us to compute the coefficients described in Proposi-
tion 4.7 for Toeplitz operators. The first step is given by the following result. We
recall that for everyx in the cone there exists a unique y such that y 2 = x,
and we denote y = x.
Theorem 4.9. Let D be a bounded symmetric domain and L (D) a radial


symbol. Then, for every > p 1 and for every N r we have


T , = h , i
Z

=C ( x) (x)(e x)p (x)b dx,
(e)

for some constant C independent of and .

Proof. We fix , and to compute as follows.
First, we apply the integral formula (3.1) to obtain
Z
h , i = c (z)| (z)|2 h(z, z)p dz
D
Z Z  X r   X r 2  X r r
X p
 
= cc g tj ej  g tj e j  h g tj e j , g tj e j dg
[0,1)r K j=1 j=1 j=1 j=1
Y r
Y
|t2j t2k |a t2b+1
j dt1 . . . dtr ,
1j<kr j=1

where we take
r
1 Y ( (j 1) a2 )
c =
n j=1 ( nr (j 1) a2 )

Note that we have restricted the integral to [0, 1)r in order to integrate over D.
Since , h are K-invariant and by Lemma 4.8 we now obtain
h , i =
Z X
r  Xr  X
r r
X p
cc
= tj ej t2j ej h tj e j , tj e j
d [0,1)r j=1 j=1 j=1 j=1
Y r
Y
|t2j t2k |a t2b+1
j dt1 . . . dtr
1j<kr j=1
10 MATTHEW DAWSON AND RAUL QUIROGA-BARRANCO

and using equation (3.3) we get

Z Xr  X r   Xr p
cc
= tj ej t2j ej e t2j ej
d [0,1)r j=1 j=1 j=1
Y r
Y
|t2j t2k |a t2b+1
j dt1 . . . dtr
1j<kr j=1
Z X
r  Xr   r
X p
cc
= xj ej xj ej e xj ej
d [0,1)r j=1 j=1 j=1
Y r
Y
|xj xk |a xbj dx1 . . . dxr
1j<kr j=1

using equation (2.1) we get

Z Xr  X r   Xr p
cc
= xj ej xj ej e xj ej
d [0,1)r j=1 j=1 j=1
X
r b Y
xj ej |xj xk |a dx1 . . . dxr
j=1 1j<kr

and then applying the definition of we obtain

Z X r   Xr p  X r b
cc
= xj ej e xj ej xj ej
d [0,1)r j=1 j=1 j=1
Z  X r  Y
s xj ej ds |xj xk |a dx1 . . . dxr .
L j=1 1j<kr

Since , and e are L-invariant, we are now in position to apply equation (3.2).
To do so, we note that integration over [0, 1)r for
r
X
x= xj ej ,
J=1

corresponds to the set (e ). Hence, we obtain

Z

h , i = C ( x) (x)(e x)p (x)b dx,
(e)

where C is a constant that is independent of and . Note that we have used

v
uX
u r r
X
t xj ej = xj ej ,
j=1 j=1

which follows from the mutual orthogonality of e1 , . . . , er . 

As a consequence of the proof of Theorem 4.9 we have the following.

 RADIAL TOEPLITZ OPERATORS ON CARTAN DOMAINS 11

Corollary 4.10. With the assumptions of Theorem 4.9 we have for every


> p 1 and N r


T , = h , i
Z Xr  X
r  X
r p  X
r b

=C xj ej xj ej (1 xj )ej xj ej
[0,1)r j=1 j=1 j=1 j=1
Y
a
|xj xk | dx1 . . . dxr ,
1j<kr

for some constant C independent of and .

A direct application of Theorem 4.9 yields the following expression of the coef-
ficients in Proposition 4.7 for radial Toeplitz operators in terms of the polynomials
(z) and j (z). We recall that for our previous notation we have (z) = r (z)
for a domain of rank r.
Theorem 4.11. Let D be a bounded symmetric domain and L (D) a


radial symbol. Then, for every > p 1 and for every N r the coefficients
c (T ) from Proposition 4.7 satisfy
Z


 ( x) (x)(e x)p (x)b dx
T , (e)
c (T ) = = Z
h , i
(x)(e x)p (x)b dx
(e)
Z

( x)1 (x)1 2 2 (x)2 3 . . . r (x)r +b r (e x)p dx
(e)
= Z .
1 (x)1 2 2 (x)2 3 . . . r (x)r +b r (e x)p dx
(e)

We can now apply Corollary 4.10 and use (2.1) to write down the formulas from
Theorem 4.11 in terms of Lebesgue integrals over [0, 1)r to obtain the following
result.
Theorem 4.12. Let D be a bounded symmetric domain and L (D) a


radial symbol. Then, for every > p 1 and for every N r the coefficients
c (T ) from Proposition 4.7 satisfy


T ,
c (T ) = =
h , i
Z X r Y
r Yr Y
+b
xj ej xj j (1 xj )p |xj xk |a dx1 . . . dxr
[0,1)r j=1 j=1 j=1 1j<kr
Z r r .
Y +b
Y Y
p a
xj j (1 xj ) |xj xk | dx1 . . . dxr
[0,1)r j=1 j=1 1j<kr

5. Radial Toeplitz operators on the Bergman spaces

All the data associated to the domain D involved in Theorem 4.12 is readily
available in the references (see [5, 18]). We collect this information and provide
specific formulas for all irreducible bounded symmetric domains. Note that all the
12 MATTHEW DAWSON AND RAUL QUIROGA-BARRANCO

information for the classical cases are explicitly contained in the references, but not
so in the exceptional cases. For the latter we provide the required arguments.

I
5.1. Type I. Up to finite index covering maps the domain Dm,n is given by
the groups
G = SU(m, n), K = S(U(m) U(n)).
For simplicity and without loss of generality we will assume that m n. The
elements of K are matrices of the form
 
A 0
,
0 B
where A U(m), B U(n) and det(A) det(B) = 1.
The domain in this case is given by
I
Dm,n = {Z Mmn (C) | ZZ < Im },
where the K-action has the expression
 
A 0
Z = AZB 1 ,
0 B
I
where A, B are as above and Z Dm,n . And we have
I
dim Dm,n = mn, r = m, a = 2, b = n m, p = n + m.
A maximal collection of mutually orthogonal primitives is given by the m n
matrices E1 , . . . , Em , where Ej has 1 at the (j, j)-th entry and 0 elsewhere. Hence,
we have
E = (Im , 0),
where 0 denotes here a m (n m) matrix. Proposition 4.5 states that for every
I
Z Dm,n there exist A U(m), B U(n) and a diagonal matrix D of size m m
with entries in [0, 1) such that
Z = A(D, 0)B 1 .


It follows that for every > n + m 1, bounded radial symbol and N m
we have


T ,
c (T ) = =
h , i
Z Ym m
Y Y
+nm
(D( x), 0) xj j (1 xj )mn |xj xk |2 dx1 . . . dxm
[0,1)m j=1 j=1 1j<km
Z m m ,
Y +nm
Y Y
xj j (1 xj )mn |xj xk |2 dx1 . . . dxm
[0,1)m j=1 j=1 1j<km

where D( x) denotes the m m diagonal matrix whose entries along the diagonal

are x1 , . . . , xm .
If we assume that m = 1, then the domain is the unit ball Bn in Cn and the
previous formula reduces to Theorem 6.2 from [14].
 RADIAL TOEPLITZ OPERATORS ON CARTAN DOMAINS 13

II
5.2. Type II. Up to finite index covering maps the domain Dm is given by
the groups
!

A B 
G = SO (2m) = A, B Mm (C) SU(m, m)
B A
!

A 0 
K = U(n) = A U(m) .
0 A

The domain is give by

II
Dm = {Z Mm (C) | ZZ < Im , Z t = Z},

with K-action
A Z = AZAt ,
II
where A U(n) and Z Dm . The properties of this domain depend on the parity
of m. Hence, we will write m = 2n + where {0, 1}. With this notation we now
have
II
dim Dm = n(2n + 2 1), r = n, a = 4, b = 2, p = 4n + 2 2.

For every j = 1, . . . , n let us denote by Ej Mm (C) the block diagonal matrix

of the form

0 0
. . ..
. ..
. .

0 1 ,
0
1 0
. .. ..
.. . .

where the only non-zero block is the (j, j)-th one. In this description the blocks are
2 2 from the upper left corner and the j-th diagonal block has the pictured entry.
Note that for = 1, we have zero blocks of size 2 1 at the far right, of size 1 2 at
the lower bottom and a single 0 at the lower right corner. Hence, E1 , . . . , En is a
maximal collection of mutually orthogonal primitives. From this it is easy to write
down E.
II
For this case, Proposition 4.5 states that for every Z Dm there exist A
U(n), t1 , . . . , tn [0, 1) such that

Z = AD(t)At ,

where D(t) = D(t1 , . . . , tn ) = Diag(d(t1 ), . . . , d(tj ), . . . ) is a m m matrix that is

block diagonal with 2 2 blocks along the diagonal starting from the upper left
corner and given by
 
0 tj
d(tj ) = .
tj 0
As before, we have to complete the matrix D(t) with some smaller zero blocks for
the case = 1.
14 MATTHEW DAWSON AND RAUL QUIROGA-BARRANCO

It follows that for > 4n + 2 3, bounded radial symbol and N n we
have


T ,
c (T ) = =
h , i
Z Y n n
Y Y
+2
(D( x)) xj j (1 xj )4n2+2 |xj xk |4 dx1 . . . dxn
[0,1)n j=1 j=1 1j<kn
Z n n ,
Y +2
Y Y
4n2+2 4
xj j (1 xj ) |xj xk | dx1 . . . dxn
[0,1)n j=1 j=1 1j<kn

where the block diagonal matrix D( x) = D( x1 , . . . , xn ) is given as before.

5.3. Type III. Up to finite index covering maps the domain DnIII is given by
the groups
G = Sp(n, R),
!

A 0 
K = U(n) = A U(n) .
0 A

The domain is naturally identified is given by

DnIII = {Z Mn (C) | ZZ < In , Z t = Z},
with K-action
A Z = AZAt ,
where A U(n) and Z DnIII . For this domain we have
n(n + 1)
dim DnIII = , r = n, a = 1, b = 0, p = n + 1.
2
A maximal collection of mutually orthogonal primitives is given by E1 , . . . , En ,
the canonical basis for the space of diagonal matrices in Mn (C). In particular, we
have E = In . Proposition 4.5 states that for every Z DnIII there exist A U(n)
and a diagonal matrix D with entries in [0, 1) such that
Z = ADAt .


It follows that for every > n, a bounded radial symbol and N n we
have


T ,
c (T ) = =
h , i
Z Yn n
Y Y

(D( x)) xj j (1 xj )n1 |xj xk | dx1 . . . dxn
[0,1)n j=1 j=1 1j<kr
Z n n ,
Y
Y Y
xj j (1 xj )n1 |xj xk | dx1 . . . dxm
n
[0,1) j=1 j=1 1j<km

where D( x) denotes the n n diagonal matrix whose entries along the diagonal

are x1 , . . . , xn .
 RADIAL TOEPLITZ OPERATORS ON CARTAN DOMAINS 15

5.4. Type IV. The groups associated to the domain DnIV can be taken to be
G = SO(n, 2)
K = SO(n) SO(2).
For this domain we have the realization
n  1 X n o
  
DnIV = z Cn |z|2 < 2, |z|2 < 1 +  zj2 
2 j=1

with the K-action  

A 0
z = ei Az,
0 B
where z DnIV , A SO(n) and B SO(2) has the form
 
cos() sin()
B= .
sin() cos()
This domain has the following data
dim DnIV = n, r = 2, a = n 1, b = 0, p = n.
A maximal collection of mutually orthogonal primitives is given by
1 i 
e1 = , , 0, . . . , 0
2 2
1 i 
e2 = , , 0, . . . , 0 ,
2 2
and so we have e = (1, 0, . . . , 0). Proposition 4.5 says that for every z DnIV there
is A SO(n), R and t1 , t2 [0, 1) such that
ei Az = t1 e1 + t2 e2 .


We now have that for every > n 1, a bounded radial symbol and N 2
we have


T ,
c (T ) = =
h , i
Z

( x1 e1 + x2 e2 )x 1 2
1 x2 (1 x1 )
n
(1 x2 )n |x1 x2 |n1 dx1 dx2
[0,1)2
Z .
1 2 n n n1
x1 x2 (1 x1 ) (1 x2 ) |x1 x2 | dx1 dx2
[0,1)2

5.5. Exceptional type V. In the notation of [5], the symmetric pair of Lie
algebras that realize this domain is (e6(14) , so(10) + R), also known as E III.
The properties of this domain are better described in terms of the octonians O.
We recall that O is an 8-dimensional real composition algebra whose basic properties
can be found in the literature (see [16]). We consider the complexified octonions
OC = O R C and the 16-dimensional complex vector space O2C . According to [12],
the latter carries a Jordan pair structure that such that
y t x),
{xyx} = x(e
for every x, y O2C . Here, ye is computed component-wise and comes from the the
involution on OC obtained by complexifying the canonical conjugation on O.
16 MATTHEW DAWSON AND RAUL QUIROGA-BARRANCO

The exceptional domain DV has O2C as associated Jordan pair (see [12]) and
so it is a bounded domain in this complex vector space. The data for this domain
is the following
dim DV = 16, r = 2, a = 6, b = 4, p = 12.
Furthermore, using the Jordan pair structure described it is easy to check that a
maximal collection of mutually orthogonal primitives is given by
e1 = (1, 0), e2 = (0, 1),
and so e = (1, 1). In this case, Proposition 4.5 states that for every (a, b) DV
there exists g K (where K SO(10) T) and (t1 , t2 ) [0, 1)2 such that
(a, b) = g(t1 , t2 ).
With respect to Toeplitz operators we now conclude that for every > 11,


a bounded radial symbol and N 2 we have


T ,
c (T ) = =
h , i
Z
1 +4 2 +4
( x1 , x2 )x 1 21 (1 x1 )12 (1 x2 )12 |x1 x2 |6 dx1 dx2
[0,1)2
Z .
1 +4 2 +4
x
1 2 1 (1 x1 )12
(1 x2 )12
|x1 x 2 | 6
dx1 dx2
[0,1)2

5.6. Exceptional type VI. In the notation of [5], the symmetric pair of Lie
algebras that realize this domain is (e7(25) , e6 + R) which is also known as E V II.
Let us consider the 27-dimensional complex vector space H3 (OC ) of Hermitian
3 3 matrices with entries in OC . By Hermitian we mean to satisfy
et = x.
x
In particular, the elements of H3 (OC ) are of the form

z1 a b
e
a z2 c ,
eb e c z3
where z1 , z2 , z3 C and a, b, c OC . For x, y H3 (OC ) denote the commutative
product
x y = xy + yx.
Then, H3 (OC ) is a Jordan pair for the operation
1
{xyx} = (x (x y) x2 y)
2
for every x, y H3 (OC ), where x2 is just the square with the respect to product of
matrices. As shown in [12], this is the Jordan pair associated to the domain DV I .
In particular, DV I has a realization as a circled bounded domain in H3 (OC ). The
data for this domain is given by
dim DV I = 27, r = 3, a = 8, b = 0, p = 26.
From the previous Jordan pair structure it is easy to see that a maximal collection
of mutually orthogonal primitives is
e1 = E1,2 + E2,1 , e2 = E1,3 + E3,1 , e3 = E2,3 + E3,2 ,
 RADIAL TOEPLITZ OPERATORS ON CARTAN DOMAINS 17

where as usual Ej,k denotes the 33 matrix with all entries 0 except at the position
(j, k) where it is 1. For this case, Proposition 4.5 implies that for every x DV I
there are g K (where K E6 T) and (t1 , t2 , t3 ) [0, 1)3 such that

0 t1 t2
x = g(t1 e1 + t2 e2 + t3 e3 ) = g t1 0 t3 .
t2 t3 0
And for the Toeplitz operators we have that for every > 25, a bounded


radial symbol and N 3 we have


T ,
c (T ) = =
h , i
Z X 3 Y3 3
Y Y

xj ej xj j (1 xj )26 |xj xk |8 dx1 dx2 dx3
[0,1)3 j=1 j=1 j=1 1j<k3
Z 3 3
.
Y
Y Y
26 8
xj j (1 xj ) |xj xk | dx1 dx2 dxr
[0,1)3 j=1 j=1 1j<k3

References
1. M. Dawson, G. Olafsson and R. Quiroga-Barranco, Commuting Toeplitz operators on
bounded symmetric domains and multiplicity-free restrictions of holomorphic discrete se-
ries, Journal of Functional Analysis 268 (2015), no. 7, 17111732.
2. J. Faraut and A. Koranyi, Function spaces and reproducing kernels on bounded symmetric
domains. J. Funct. Anal. 88 (1990), no. 1, 64-89.
3. , Analysis on Symmetric Cones, Oxford Science Publications, Clarendon Press, Ox-
ford, 1994.
4. R. Goodman and N. R. Wallach, Symmetry, representations, and invariants. Graduate Texts
in Mathematics, 255. Springer, Dordrecht, 2009.
5. S. Helgason, Differential geometry, Lie groups and symmetric spaces, Graduate Studies in
Mathematics, 34. American Mathematical Society, Providence, RI, 2001.
6. , Groups and Geometric Analysis, Mathematical Surveys and Monographs, 83. Amer-
ican Mathematical Society, Providence, RI, 2000.
7. L. K. Hua, Harmonic analysis of functions of several complex variables in the classical do-
mains. Translated from the Russian, which was a translation of the Chinese original, by Leo
Ebner and Adam Koranyi. With a foreword by M. I. Graev. Reprint of the 1963 edition.
Translations of Mathematical Monographs, 6. American Mathematical Society, Providence,
R.I., 1979.
8. K. D. Johnson, On a ring of invariant polynomials on a Hermitian symmetric space, J.
Algebra 67 (1980), no. 1, 7281.
9. T. Kobayashi, Multiplicity-free theorems of the restrictions of unitary highest-weight modules
with respect to reductive symmetric pairs, Representation Theory and Automorphic Forms,
Progress in Mathematics 255 (2008), 45109.
10. A. Koranyi, Analytic invariants of bounded symmetric domains, Proc. Amer. Math. Soc. 19
(1968), 279284.
11. A. Koranyi and Stephen Vagi, Rational inner functions on bounded symmetric domains,
Trans. Amer. Math. Soc. 254 (1979), 179193.
12. O. Loos, Bounded symmetric domains and Jordan pairs, The University of California, Irvine,
1977.
13. G. Olafsson and B. rsted, Generalizations of the Bargmann transform, in: Lie Theory and
Its Applications in Physics, Clausthal, 1995, World Sci. Publ., River Edge, NJ, 1996, pp.
314.
14. R. Quiroga-Barranco, Separately radial and radial Toeplitz operators on the unit ball and
representation theory, Boletin de la Sociedad Matematica Mexicana 22 (2016), no. 2, 605623
18 MATTHEW DAWSON AND RAUL QUIROGA-BARRANCO

15. W. Schmid, Die Randwerte holomorpher Funktionen auf hermitesch symmetrischen

Raumen, Invent. Math. 9 (1969), 6180.
16. T. A. Springer and F. D. Veldkamp, Octonions, Jordan algebras and exceptional groups.
Springer Monographs in Mathematics. Springer-Verlag, Berlin, 2000.
17. H. Upmeier, Toeplitz C -Algebras on Bounded Symmetric Domains, Ann. Math. 119 (1984),
no. 3, 549576.
18. H. Upmeier, Toeplitz operators and index theory in several complex variables, Operator
Theory: Advances and Applications, 81. Birkhauser Verlag, Basel, 1996. x+481 pp.
19. N. R. Wallach, The analytic continuation of the discrete series I, Transactions of the Amer-
ican Mathematical Society 251 (1979), 117.

Centro de Investigacion en Matematicas, Merida, Mexico

E-mail address: matthew.dawson@cimat.mx

Centro de Investigacion en Matematicas, Guanajuato, Mexico

E-mail address: quiroga@cimat.mx