C -ALGEBRAS ASSOCIATED WITH HILBERT C -QUAD

MODULES OF FINITE TYPE

KENGO MATSUMOTO

Abstract. A Hilbert C -quad module of nite type has a multi structure of

Hilbert C -bimodules with two nite bases. We will construct a C -algebra
from a Hilbert C -quad module of nite type and prove its universality subject
to certain relations among generators. Some examples of the C -algebras from
Hilbert C -quad modules of nite type will be presented.

1. Introduction
G. RobertsonT. Steger [23] have initiated a certain study of higher dimensional
analogue of CuntzKrieger algebras from the view point of tiling systems of 2-
dimensional plane. After their work, A. KumjianD. Pask [11] have generalized
their construction to introduce the notion of higher rank graphs and its C -algebras.
Since then, there have been many studies on these C -algebras by many authors
(see for example [6], [7], [11], [22], [18], [23], etc.).
In [12], the author has introduced a notion of C -symbolic dynamical system,
which is a generalization of a nite labeled graph, a -graph system and an automor-
phism of a unital C -algebra. It is denoted by (A, , ) and consists of a nite family

{ }P
of endomorphisms of a unital C -algebra A such that (ZA ) ZA ,
and (1) 1 where ZA denotes the center of A. It provides a subshift
over and a Hilbert C -bimodule HA
over A which gives rise to a C -algebra O
as a Cuntz-Pimsner algebra ([12], cf. [8], [16], [21]). In [13] and [14], the author
has extended the notion of C -symbolic dynamical system to C -textile dynamical
system which is a higher dimensional analogue of C -symbolic dynamical system.
A C -textile dynamical system (A, , , , , ) consists of two C -symbolic dy-
namical systems (A, , ) and (A, , ) with common unital C -algebra A and
commutation relations between the endomorphisms , and a , a .
A C -textile dynamical system provides a two-dimensional subshift and a multi
structure of Hilbert C -bimodules that has multi right actions and multi left ac-
tions and multi inner products. Such a multi structure of Hilbert C -bimodule is
called a Hilbert C -quad module. In [14], the author has introduced a C -algebra
associated with the Hilbert C -quad module of C -textile dynamical system. It
is generated by the quotient images of creation operators on two-dimensional ana-
logue of Fock Hilbert module by module maps of compact operators. As a result,
the C -algebra has been proved to have a universal property subject to certain op-
erator relations of generators encoded by structure of C -textile dynamical system
([14]).
1991 Mathematics Subject Classication. Primary 46L35; Secondary 46L08.
Key words and phrases. C -algebras, Hilbert C -modules, symbolic dynamics, Cuntz algebras,
Cuntz-Krieger algebras.
1
 In this paper, we will generalize the construction of the C -algebras of Hilbert
C -quad modules of C -textile dynamical systems. Let A, B1 , B2 be unital C -

algebras. Assume that A has unital embeddings into both B1 and B2 . A Hilbert
C -quad module H over (A; B1 , B2 ) is a Hilbert C -bimodule over A with A-valued
right inner product | A which has a multi structure of Hilbert C -bimodules over
Bi with right actions i of Bi and left actions i of Bi and Bi -valued inner products
| Bi for i = 1, 2 satisfying certain compatibility conditions. A Hilbert C -quad
module H is said to be of nite type if there exist a nite basis {u1 , . . . , uM } of
H as a Hilbert C -right module over B1 and a nite basis {v1 , . . . , vN } of H as a
Hilbert C -right module over B2 such that
X
M X
N
ui | 2 ( | B2 )ui B1 = vk | 1 ( | B1 )vk B2 = | A
i=1 k=1
for , H (see [26] for the original denition of nite basis of Hilbert module).
For a Hilbert C -quad module, we will construct a Fock space F (H) from H, which
is a 2-dimensional analogue to the ordinary Fock space of Hilbert C -bimodules (cf.
[10], [21]). We will then dene two kinds of creation operators s , t for H on
F (H). The C -algebra on F (H) generated by them is denoted by TF (H) and called
the Toeplitz quad module algebra. We then dene the C -algebra OF (H) associated
with the Hilbert C -quad module H by the quotient C -algebra of TF (H) by the
ideal generated by the nite rank operators. We will then prove that the C -
algebra OF (H) for a C -quad module H of nite type has a universal property in
the following way.
Theorem 1.1 (Theorem 5.17). Let H be a Hilbert C -quad module over (A; B1 , B2 )
of nite type with a nite basis {u1 , . . . , uM } of H as a Hilbert C -right module over
B1 and a nite basis {v1 , . . . , vN } of H as a Hilbert C -right module over B2 . Then
the C -algebra OF (H) generated by the quotients [s ], [t ] of the creation operators
s , t for H on the Fock spaces F (H) is canonically isomorphic to the universal
C -algebra OH generated by operators S1 , . . . , SM , T1 , . . . , TN and elements z
B1 , w B2 subject to the relations:
X
M X
N
Si Si + Tk Tk = 1, Sj Tl = 0,
i=1 k=1
Si Sj = ui | uj B1 , Tk Tl = vk | vl B2 ,
X
M X
N
zSj = Si ui | 1 (z)uj B1 , zTl = Tk vk | 1 (z)vl B2 ,
i=1 k=1
X
M X
N
wSj = Si ui | 2 (w)uj B1 , wTl = Tk vk | 2 (w)vl B2
i=1 k=1
for z B1 , w B2 , i, j = 1, . . . , M, k, l = 1, . . . , N .
The eight relations of the operators above are called the relations (H). As a
corollary we have
Corollary 1.2 (Corollary 5.18). For a C -quad module H of nite type, the uni-
versal C -algebra OH generated by operators S1 , . . . , SM , T1 , . . . , TN and elements
z B1 , w B2 subject to the relations (H) does not depend on the choice of the
nite bases {u1 , . . . , uM } and {v1 , . . . , vN }.
2
 The paper is organized in the following way. In Section 2, we will dene Hilbert
C -quad module and present some basic properties. In Section 3, we will dene a
C -algebra OF (H) from Hilbert C -quad module H of general type by using creation
operators on Fock Hilbert C -quad module. In Section 4, we will study algebraic
structure of the C -algebra OF (H) for a Hilbert C -quad module H of nite type.
In Section 5, we will prove, as a main result of the paper, that the C -algebra OF (H)
has the universal property stated as in Theorem 1.1. A sterategy to prove Theorem
1.1 is to show that the C -algebra OF (H) is regarded as a Cuntz-Pimsner algebra
for a Hilbert C -bimodule over the C -algebra generated by 1 (B1 ) and 2 (B2 ). We
will then prove the gauge invariant universarity of the C -algebra (Theorem 5.16).
In Section 6, we will present K-theory formulae for the C -algebra OH . In Section
7, we will give examples. In Section 8, we will formulate higher dimensional anlogue
of our situations and state a generalized proposition of Theorem 1.1 without proof.
Throughout the paper, we will denote by Z+ the set of nonnegative integers and
by N the set of positive integers.

2. Hilbert C -quad modules

Throughout the paper we x three unital C -algebras A, B1 , B2 such that A
B1 , A B2 with common units. We assume that there exists a right action i of
A on Bi so that
bi i (a) Bi for bi Bi , a A, i = 1, 2,
which satises
bi i (a) bi a, bi i (aa ) = bi i (a)i (a )
for bi Bi , a, a A, i = 1, 2. Hence Bi is a right A-module through i for i = 1, 2.
Suppose that H is a Hilbert C -bimodule over A, which has a right action of A,
an A-valued right inner product | A and a -homomorphism A from A to the
algebra of all bounded adjointable right A-module maps LA (H) satisfying
(i) | A is linear in the second variable.
(ii) | aA = | A a for , H, a A.
(iii) | A = | A for , H.
(iv) | A 0, and | A = 0 if and only if = 0.
A Hilbert C -bimodule H over A is called a Hilbert C -quad module over (A; B1 , B2 )
if H has a further structure of a Hilbert C -bimodule over Bi for each i = 1, 2 with
right action i of Bi and left action i of Bi and Bi -valued right inner product
| Bi such that for z B1 , w B2
1 (z), 2 (w) LA (H)
and
[1 (z)]2 (w) = 1 (z)[2 (w)], [2 (w)]1 (z) = 2 (w)[1 (z)],
(2.1)
1 (z1 (a)) = [1 (z)]a, 2 (w2 (a)) = [2 (w)]a, (2.2)
for H, z B1 , w B2 , a A and
A (a) = 1 (a) = 2 (a) for a A (2.3)
where A is regarded as a subalgebra of Bi . The left action i of Bi on H means
that i (bi ) for bi Bi is a bounded adjointable operator with respect to the inner
3
product | Bi for each i = 1, 2. The operator i (bi ) for bi Bi is also adjointable
with respect to the inner product | A . We assume that the adjoint of i (bi )
with respect to the inner product | Bi coincides with the adjoint of i (bi ) with
respect to the inner product | A . Both of them coincide with i (bi ). We assume
that the left actions i of Bi on H for i = 1, 2 are faithful. We require the following
compatibility conditions between the right A-module structure of H and the right
A-module structure of Bi through i :

| aBi = | Bi i (a) for , H, a A, i = 1, 2. (2.4)

We further assume that H is a full Hilbert C -bimodule with respect to the

three inner products | A , | B1 , | B2 for each. This means that the C -
algebras generated by elements { | A | , H}, { | B1 | , H} and
{ | B2 | , H} coincide with A, B1 and B2 respectively.
1
For a vector H, denote by A , B1 , B2 the norms | A 2 , |
1 1
B1 2 , | B2 2 induced by the right inner products respectively. By denition,
H is complete under the above three norms for each.
Denition.
(i) A Hilbert C -quad module H over (A; B1 , B2 ) is said to be of general type if
there exists a faithful completely positive map i : Bi A for i = 1, 2 such that

i (bi i (a)) = i (bi )a for bi Bi , a A, (2.5)

i ( | Bi ) = | A , for , H, i = 1, 2. (2.6)

(ii) A Hilbert C -quad module H over (A; B1 , B2 ) is said to be of nite type if

there exist a nite basis {u1 , . . . , uM } of H as a right Hilbert B1 -module and a nite
basis {v1 , . . . , vN } of H as a right Hilbert B2 -module, that is,

X
M X
N
ui 1 (ui | B1 ) = vk 2 (vk | B2 ) = , H (2.7)
i=1 k=1

such that

ui | 2 (w)uj B1 A, i, j = 1, . . . , M, (2.8)
vk | 1 (z)vl B2 A, k, l = 1, . . . , N (2.9)

for w B2 , z B1 and

X
M
ui | 2 ( | B2 )ui B1 = | A , (2.10)
i=1
X
N
vk | 1 ( | B1 )vk B2 = | A (2.11)
k=1

for all , H. Following [26], {u1 , . . . , uM } and {v1 , . . . , vN } are called nite
bases of H respectively.
(iii) A Hilbert C -quad module H over (A; B1 , B2 ) is said to be of strongly nite
type if it is of nite type and there exist a nite basis {e1 , . . . , eM } of B1 as a right
A-module through 1 and a nite basis {f1 , . . . , fN } of B2 as a right A-module
4
through 2 2 . This means that the following equalities hold:

X
M
z= ej 1 (1 (ej z)), z B1 , (2.12)
j=1

X
N
w= fl 2 (2 (fl w)), w B2 . (2.13)
l=1

We note that for a Hilbert C -quad module of general type, the conditions (2.6)
imply
| A i (1) | Bi , H.
1
Put Ci = i (1) 2 > 0 so that A Ci Bi . Hence the identity operators
from the Banach spaces (H, Bi ) to (H, A ) are bounded linear maps. By the
inverse mapping theorem, there exist constants Ci such that Bi Ci A for
H. Therefore the three norms A , Bi , i = 1, 2, induced by the three inner
products | A , | Bi , i = 1, 2 on H are equivalent to each other.
Lemma 2.1. Let H be a Hilbert C -quad module H over (A; B1 , B2 ). If H is of
nite type, then it is of general type.
Proof. Suppose that H is of nite type with nite bases {u1 , . . . , uM } of H as a
right Hilbert B1 -module and {v1 , . . . , vN } of H as a right Hilbert B2 -module as
above. We put
X
N
1 (z) = vk | 1 (z)vk B2 A, z B1 ,
k=1
X
M
2 (w) = ui | 2 (w)ui B1 A, w B2 .
i=1

They give rise to faithful completely positive maps i : Bi A, i = 1, 2. The

equalities (2.10) (2.11) imply that
i ( | Bi ) = | A , for , H, i = 1, 2. (2.14)
It then follows that
i ( | Bi i (a)) = i ( | aBi ) = | A a = i ( | Bi )a, a A.
Since H is full, the equalities (2.5) hold.

Lemma 2.2. Suppose that a Hilbert C -quad module H of nite type is of strongly
nite type with a nite basis {e1 , . . . , eM } of B1 as a right A-module through 1
1 and a nite basis {f1 , . . . , fN } of B2 as a right A-module through 2 2 .
Let {u1 , . . . , uM } and {v1 , . . . , vN } be nite bases of H satisfying (2.7). Then two
families {ui 1 (ej ) | i = 1, . . . , M, j = 1, . . . , M } and {vk 2 (fl ) | k = 1, . . . , N, l =
1, . . . , N } of H form bases of H as right A-modules respectively.
Proof. For H, by the equalities
X
M X
M
= ui 1 (ui | B1 ), ui | B1 = ej 1 (1 (ej ui | B1 )),
i=1 j=1

5
it follows that

X
M XM
= ui 1 ( ej 1 (1 (ej ui | B1 )))
i=1 j=1
M
M X
X
= ui 1 (ej ) 1 (ej ui | B1 )
i=1 j=1

X
M X
M
= ui 1 (ej ) 1 (ui 1 (ej ) | B1 )
i=1 j=1

X
M X
M
= ui 1 (ej ) ui 1 (ej ) | A .
i=1 j=1

We similarly have

X
N X
N
= vk 2 (fl ) vk 2 (fl ) | A .
k=1 l=1

We present some examples.

Examples.
1. Let , be automorphisms of a unital C -algebra A satisfying = .
Dene right actions i of A on Bi by
b1 1 (a) = b1 (a), b2 2 (a) = b2 (a)
for bi Bi , a A. We set B1 = B2 = A. We put H, = A and equip it
with Hilbert C -quad module structure over (A; A, A) in the following way. For
= x, = x H, = A, a A, z B1 = A, w B2 = A, dene the right
A-module structure and the right A-valued inner product | A by
a = xa, | A = x x .
Dene the right actions i of Bi with right Bi -valued inner products | Bi and
the left actions i of Bi by setting
1 (z) = x(z), 2 (w) = x(w),
1
| B1 = (x x ), | B2 = 1 (x x ),
1 (z) = ((z))x, 2 (w) = ((w))x.
It is straightforward to see that H, is a Hilbert C -quad module over (A; A, A)
of strongly nite type.
2. We x natural numbers 1 < N, M N. Consider nite dimensional com-
mutative C -algebras A = C, B1 = CN , B2 = CM . The right actions i of A on
Bi are naturally dened as right multiplications of C. The algebras B1 , B2 have
the ordinary product structure and the inner product structure which we denote
by | N and | M respectively. Let us denote by HM,N the tensor product
CM CN . Dene the right actions i of Bi with Bi -valued right inner products
6
 | Bi and the left actions i of Bi on HM,N = CM CN for i = 1, 2 by setting

( )1 (z) = ( z), ( )2 (w) = ( w) ,

| B1 = | M B1 ,

| B2 = | N B2 ,
1 (z)( ) = (z ), 2 (w)( ) = (w )

for z B1 , , CN , w B2 , , CM . Let ei , i = 1, . . . , M and fk , k = 1, . . . , N

be the standard basis of CM and that of CN respectively. Put the nite bases

ui = ei 1 HM,N , i = 1, . . . , M,
vk = 1 fk HM,N , i = 1, . . . , N.

It is straightforward to see that HM,N is a Hilbert C -quad module over (C; CN , CM )

of strongly nite type.

3. Let (A, , , , , ) be a C -textile dynamical system which means that for

j , l endomorphisms j , l of A are given with commutation relations j
= k i if (l, j) = (i, k). In [14], a Hilbert C -quad module H, over (A; B1 , B2 )
from (A, , , , , ) is constructed (see [14] for its detail construction). The
two triplets (A, , ) and (A, , ) are C -symbolic dynamical systems ([12]),
that yield C -algebras O and O respectively. The C -algebras B1 and B2 are
dened as the C -subalgebra of O generated by elements Tj yTj , j , y A
and that of O generated by Sk ySk , k , y A respectively. Dene the maps
i : A Bi , i = 1, 2 by
X X
1 (a) = Tj aTj , 2 (a) = Sl aSl , aA
j l

which yield the right actions of A on Bi , i = 1, 2. Dene the maps i : Bi A, i =

1, 2 by
X X
1 (z) = Tj zTj , 2 (w) = Sl wSl , z B1 , w B2 .
j l
P
Put ej = Tj Tj B1 , j . Let z = j Tj zj Tj be an element of B1 for zj A
with Tj Tj zj Tj Tj = zj . As 1 (ej z) = 1 (Tj zj Tj ) = zj , one sees
X X X
z= Tj Tj Tj zj Tj = Tj Tj 1 (zj ) = ej 1 (1 (ej z)).
j j j

We similarly have by putting fl = Sl Sl B2 ,

X
w= fl 2 (2 (fl w)) for w B2 .
l

We see that H, is a Hilbert C -quad module of strongly nite type. In particular,

two nonnegative commuting matrices A, B with a specication coming from the
equality AB = BA yield a C -textile dynamical system and hence a Hilbert C -
quad module of strongly nite type, which are studied in [15].
7
 3. Fock Hilbert C -quad modules and creation operators
In this section, we will construct a C -algebra from a Hilbert C -quad mod-
ule H of general type by using two kinds of creation operators on Fock space of
Hilbert C -quad module. We rst consider relative tensor products of Hilbert C -
quad modules and then introduce Fock space of Hilbert C -quad modules which
is a two-dimensional analogue of Fock space of Hilbert C -bimodules. We x a
Hilbert C -quad module H over (A; B1 , B2 ) of general type as in the preceding
section. The Hilbert C -quad module H is originally a Hilbert C -right module
over A with A-valued inner product | A . It has two other structure of Hilbert
C -bimodules, the Hilbert C -bimodule (1 , H, 1 ) over B1 and the Hilbert C -
bimodule (2 , H, 2 ) over B2 where i is a left action of Bi on H and i is a right
action of Bi on H with Bi -valued right inner product | Bi for each i = 1, 2. This
situation is written as in the gure:

B2


2 y

1 1
B1 H B1
x

2

B2

We will dene two kinds of relative tensor products

H B1 H, H B2 H

as Hilbert C -quad modules over (A; B1 , B2 ). The latter one should be written
vertically as
H
B2
H

rather than horizontally H B2 H. The rst relative tensor product H B1 H is

dened as the relative tensor product as Hilbert C -modules over B1 , where the
left H is a right B1 -module through 1 and the right H is a left B1 -module through
1 . It has a right B1 -valued inner product and a right B2 -valued inner product
dened by

B1 | B1 B1 := | B1 ( | B1 ) B1 ,
B1 | B1 B2 := | B1 ( | B1 ) B2

respectively. It has two right actions, id 1 from B1 and id 2 from B2 . It also

has two left actions, 1 id from B1 and 2 id from B2 . By these operations
H B1 H is a Hilbert C -bimodule over B1 as well as a Hilbert C -bimodule over
B2 . It also has a right A-valued inner product dened by

B1 | B1 A := 1 ( B1 | B1 B1 )(= 2 ( B1 | B1 B2 )),
8
a right A-action id a for a A and a left A-action A id. By these structure
H B1 H is a Hilbert C -quad module over (A; B1 , B2 ).

B2


2 idy

1 id id1
B1 H B1 H B1
x

id2

B2

We denote the above operations 1 id, 2 id, id1 , id2 still by 1 , 2 , 1 , 2

respectively. Similarly we consider the other relative tensor product HB2 H dened
by the relative tensor product as Hilbert C -modules over B2 , where the left H is
a right B2 -module through 2 and the right H is a left B2 -module through 2 . By
a symmetric discussion to the above, H B2 H is a Hilbert C -quad module over
(A; B1 , B2 ). The following lemma is routine.

Lemma 3.1. Let Hi = H, i = 1, 2, 3. The correspondences

(1 B1 2 ) B2 3 (H1 B1 H2 ) B2 H3 1 B1 (2 B2 3 ) H1 B1 (H2 B2 H3 ),
(1 B2 2 ) B1 3 (H1 B2 H2 ) B1 H3 1 B2 (2 B1 3 ) H1 B2 (H2 B1 H3 )

yield isomorphisms of Hilbert C -quad modules respectively.

We write the isomorphism class of the former Hilbert C -quad modules as H1 B1

H2 B2 H3 and that of the latter ones as H1 B2 H2 B1 H3 respectively.

Note that the direct sum B1 B2 has a structure of a pre Hilbert C -right module
over A by the following operations: For b1 b2 , b1 b2 B1 B2 and a A, set

(b1 b2 )A (a) := b1 1 (a) b2 2 (a) B1 B2 ,

b1 b2 | b1 b2 A := 1 (b1 b1 ) + 2 (b2 b2 ) A.

By (2.5) the equality

b1 b2 | (b1 b2 )A (a)A = b1 b2 | b1 b2 A a

holds so that B1 B2 is a pre Hilbert C -right module over A. We denote by F0 (H)

the completion of B1 B2 by the norm induced by the inner product | A . It
has right Bi -actions i and left Bi -action i by

(b1 b2 )1 (z) = b1 z 0, (b1 b2 )2 (w) = 0 b2 w,

1 (z)(b1 b2 ) = zb1 0, 2 (w)(b1 b2 ) = 0 wb2

for b1 b2 B1 B2 , z B1 , w B2 .
We denote the relative tensor product H Bi H and elements Bi by H i H
and i respectively for i = 1, 2. Let us dene the Fock Hilbert C -quad module
as a two-dimensional analogue of the Fock space of Hilbert C -bimodules. Put
9
0 = {} and n = {(i1 , . . . , in )) | ij = 1, 2}, n = 1, 2, . . . . We set
F1 (H) = H,
F2 (H) = (H 1 H ) (H 2 H),
F3 (H) = (H 1 H 1 H) (H 1 H 2 H)
(H 2 H 1 H) (H 2 H 2 H),

Fn (H) = (i1 , ,in1 )n1 H i1 H i2 in1 H
.
as Hilbert C -bimodules over A. We will dene the Fock Hilbert C -module F (H)
by setting
F (H) :=
n=0 Fn (H)

which is the completion of the algebraic direct sum

n=0 Fn (H) of the Hilbert C -

right module over A under the norm A on n=0 Fn (H) induced by the A-valued
right inner product on
n=0 Fn (H). Then F (H) is a Hilbert C -right module over
A. It has a natural left Bi -action dened by i for i = 1, 2.
For H we dene two operators
s :Fn (H) Fn+1 (H), n = 0, 1, 2, . . . ,
t :Fn (H) Fn+1 (H), n = 0, 1, 2, . . .
by setting for n = 0,
s (b1 b2 ) = 1 (b1 ), b1 b2 B1 B2 ,
t (b1 b2 ) = 2 (b2 ), b1 b2 B1 B2 ,
and for n = 1, 2, . . . ,
s (1 1 n1 n ) = 1 1 1 n1 n ,
t (1 1 n1 n ) = 2 1 1 n1 n
for 1 1 n1 n Fn (H) with (1 , . . . , n1 ) n1 .
Lemma 3.2. For H the two operators
s :Fn (H) Fn+1 (H), n = 0, 1, 2, . . . ,
t :Fn (H) Fn+1 (H), n = 0, 1, 2, . . .
are both right A-module maps.
Proof. We will show the assertion for s . For n = 0, we have for b1 b2 B1 B2
and a A,
s ((b1 b2 )A (a)) = 1 (b1 1 (a)) = (1 (b1 ))a = (s (b1 b2 ))a.
For n = 1, 2, . . . , we have
s ((1 i1 2 i2 in1 n )a)
=s ((1 i1 2 i2 in1 (n a)) = 1 1 i1 2 i2 in1 (n a)
=( 1 1 i1 2 i2 in1 n )a = [s (1 i1 2 i2 in1 n )]a.

10
It is clear that the two operators s , t yield bounded right A-module maps on F (H)
having its adjoints with respect to the A-valued right inner product on F (H). The
operators are still denoted by s , t respectively. The adjoints of s , t : F (H)
F (H) with respect to the A-valued right inner product on F (H) map Fn+1 (H) to
Fn (H), n = 0, 1, 2, . . . .
Lemma 3.3. (i) For , H = F1 (H), we have
s = | B1 0 B1 B2 , t = 0 | B2 B1 B2 .
(ii) For H and 1 i1 2 i2 in n+1 Fn+1 (H), n = 1, 2, . . . , we have
s (1 i1 2 i2 in n+1 )
(
1 ( | 1 B1 )2 i2 in n+1 if i1 = 1,
=
0 if i1 = 2,
t (1 i1 2 i2 in n+1 )
(
0 if i1 = 1,
=
2 ( | 1 B2 )2 i2 in n+1 if i1 = 2.

Proof. We will show the assertions (i) and (ii) for s . (i) For b1 b2 B1 B2 , we
have
b1 b2 | s A = 1 (b1 ) | A = 1 (b1 | B1 ) = b1 b2 | | B1 0A
so that s = | B1 0.
(ii) For 1 j1 jn1 n Fn (H) with n = 1, 2, . . . we have
1 j1 jn1 n | s (1 i1 2 i2 in n+1 )A
= 1 1 j1 jn1 n | 1 i1 2 i2 in n+1 A
=1 ( 1 1 j1 jn1 n | 1 i1 2 i2 in n+1 B1 )
(
1 j1 jn1 n | 1 ( | 1 B1 )2 i2 in n+1 A if i1 = 1,
=
0 if i1 = 2.

Denote by i the left actions of Bi , i = 1, 2 on Fn (H) and hence on F (H) respec-
tively. They satisfy the following equalities
1 (z)(b1 b2 ) = zb1 0, 2 (w)(b1 b2 ) = 0 wb2 ,
1 (z)(1 i1 2 i2 in1 n ) = (1 (z)1 ) i1 2 i2 in1 n ,
2 (w)(1 i1 2 i2 in1 n ) = (2 (w)1 ) i1 2 i2 in1 n
for z B1 , w B2 , b1 b2 B1 B2 and 1 i1 2 i2 in1 n Fn (H). More
generally let us denote by LA (H) and LA (F (H)) the C -algebras of all bounded
adjointable right A-module maps on H and on F (H) with respect to their right
A-valued inner products respectively. For L LA (H), dene L LA (F (H)) by
L(b1 b2 ) = 0 for b1 b2 B1 B2 F0 (H),
L(1 i1 2 i2 in1 n ) = (L1 ) i1 2 i2 in1 n
for 1 i1 2 i2 in1 n Fn (H).
11
Lemma 3.4. Both the maps i : Bi LA (F (H)) for i = 1, 2 are faithful -
homomorphisms.
Proof. By assumption, the -homomorphisms i : Bi LA (H), i = 1, 2 are
faithful, so that the -homomorphisms i : Bi LA (F (H)), i = 1, 2 are both
faithful.
Lemma 3.5. For , H, z B1 , w B2 , L LA (H) and c, d C, the following
equalities hold on F (H):
sc+d = cs + ds , tc+d = ct + dt , (3.1)
sL1 (z) = Ls 1 (z), tL (w) = Lt 2 (w),
2
(3.2)

s Ls = 1 ( | LB ),
1 t Lt = 2 ( | LB2 ). (3.3)
Proof. The equalities (3.1) are obvious. We will show the equalities (3.2) and (3.3)
for s . We have for b1 b2 B1 B2
sL1 (z) (b1 b2 ) = [L1 (z)]1 (b1 ) = L[1 (zb1 )] = L[s (zb1 0)]
= L[s [1 (z)(b1 b2 )]] = [Ls 1 (z)](b1 b2 ).
For 1 i1 2 i2 in1 n Fn (H), n = 1, 2, . . . , we have
sL1 (z) (1 i1 2 i2 in1 n ) = (L1 (z)) 1 1 i1 2 i2 in1 n
= L[(1 (z)) 1 1 i1 2 i2 in1 n ]
= L[ 1 (1 (z)1 ) i1 2 i2 in1 n ]
= L[s ((1 (z)1 ) i1 2 i2 in1 n )]
= Ls 1 (z)[1 i 2 i i
1 2 n ] n1

so that sL1 (z) = Ls 1 (z) on Fn (H), n = 0, 1, . . . . Hence the equalities (3.2) hold.
For b1 b2 B1 B2 , we have
s Ls (b1 b2 ) = s (L1 (b1 )) = | L1 (b1 )B1 0
= | LB b1 0 = 1 ( | LB )(b1 b2 ).
1 1

For 1 i1 2 i2 in1 n Fn (H), we have

s Ls (1 i1 2 i2 in1 n ) = s L( 1 1 i1 2 i2 in1 n )
= (1 ( | LB1 )1 ) i1 2 i2 in1 n
= 1 ( | LB )(1 i 2 i i
1 1 n
2 n1

so that s Ls = 1 ( | LB1 ) on Fn (H) for n = 0, 1, 2, . . . . Hence the equalities

(3.3) hold.
The C -subalgebra of LA (F (H)) generated by the operators s , t for H is
denoted by TF (H) and is called the Toeplitz quad module algebra for H.
Lemma 3.6. The C -algebra TF (H) contains the operators 1 (z), 2 (w) for z
B1 , w B2 .
Proof. By (3.3) in the preceding lemma, one sees
s s = 1 ( | B1 ), t t = 2 ( | B2 ), , H.
12
Since H is a full C -quad module, the inner products | B1 , | B2 for , H
generate the C -algebras B1 , B2 respectively. Hence 1 (B1 ), 2 (B2 ) are contained
in TF (H) .
Lemma 3.7. There exists an action of R/Z = T on TF (H) such that

r (s ) = e2 1r s , r (t ) = e2 1r t , H,

r (1 (z)) = 1 (z), z B1 ,
r (1 (w)) = 2 (w), w B2
for r R/Z = T.
Proof. We will rst dene a one-parameter unitary group ur , r R/Z = T on F (H)
with respect to the right A-valued inner product as in the following way.
For n = 0 : ur : F0 (H) F0 (H) is dened by
ur (b1 b2 ) = b1 b2 for b1 b2 B1 B2 .
For n = 1, 2, . . . : ur : Fn (H) Fn (H) is dened by

1nr
ur (1 i1 2 i2 in1 n ) = e2 1 i1 2 i2 in1 n
for 1 i1 2 i2 in1 n Fn (H). We therefore have a one-parameter unitary
group ur on F (H). We then dene an automorphism r on LA (F (H)) for r R/Z
by
r (T ) = ur T ur for T LA (F (H)), r R/Z.
It then follows that for b1 b2 B1 B2

r (s )(b1 b2 ) = ur s ur (b1 b2 ) = ur ((1 (b1 ))) = e2 1r
s (b1 b2 ),
and for 1 i1 in1 n Fn (H), n = 1, 2, . . .

r (s )(1 i1 in1 n ) = e2 1nr
ur ( 1 1 i1 in1 n )

1r
= e2 s (1 i1 in1 n ).

Therefore

we conclude that r (s ) = e2 1r s on F (H) and similarly r (t ) =
e2 1r t on F (H). It is direct to see that
r (1 (z)) = 1 (z), r (1 (w)) = 2 (w), for z B1 , w B2 .
It is also obvious that r (TF (H) ) = TF (H) for r R/Z.
Denote by J(H) the C -subalgebra of LA (F (H)) generated by the elements
M
nite
LA ( Fn (H)).
n=0

The algebra J(H) is a closed two-sided ideal of LA (F (H)).

Denition. The C -algebra OF (H) associated to the Hilbert C -quad module
H of general type is dened by the quotient C -algebra of TF (H) by the ideal
TF (H) J(H).
We denote by [x] the quotient image of an element x TF (H) under the ideal
TF (H) J(H). We set the elemets of OF (H)
S = [s ], T = [t ], 1 (z) = [1 (z)], 2 (w) = [2 (w)]
for H and z B1 , w B2 . By the preceding lemmas, we have
13
Proposition 3.8. The C -algebra OF (H) is generated by the family of operators
S , T for H. It contains the operators 1 (z), 2 (w) for z B1 , w B2 . They
satisfy the following equalities
Sc+d = cS + dS , Tc+d = cT + dT , (3.4)

S1 (z )1 (z) = 1 (z )S 1 (z), T1 (z )2 (w) = 1 (z )T 2 (w),
(3.5)

S2 (w )1 (z) = 2 (w )S 1 (z), T2 (w )2 (w) = 2 (w )T 2 (w),
(3.6)
S S = 1 ( | B1 ), T T = 2 ( | B2 ) (3.7)

for , H, c, d C and z, z B1 , w, w B2 .

4. The C -algebras of Hilbert C -quad modules of finite type

In what follows, we assume that a Hilbert C -quad module H is of nite type. In
this section, we will study the C -algebra OF (H) for a Hilbert C -quad module H
of nite type. Let {u1 , . . . , uM } be a nite basis of H as a Hilbert C -right module
over B1 and {v1 , . . . , vN } a nite basis of H as a Hilbert C -right module over B2 .
Keep the notations as in the previous section. We set
si = sui for i = 1, . . . , M and tk = tv k for k = 1, . . . , N.
(4.1)
By (2.7) and Lemma 3.5 we have for H
X
M X
N
s = si 1 (ui | B1 ), t = tk 2 (vk | B2 ). (4.2)
i=1 k=1
P
Let Pn be the projection on F (H) onto Fn (H) for n = 0, 1, . . . so that n=1 Pn = 1
on F (H).
Lemma 4.1. For , H, we have
(i) s t Pn = 0 for n = 1, 2, . . . and hence s t = s t P0 .
(ii) t s Pn = 0 for n = 1, 2, . . . and hence t s = t s P0 .
Proof. (i) For n = 1, 2, . . . , we have
s t (1 i1 in1 n ) = s ( 2 1 i1 in1 n ) = 0.
(ii) is smilarly shwon to (i).
Dene two projections on F (H) by

M X
Ps = The projection onto H 1 H i1 H i2 in H,
n=0 (i1 , ,in )n

M X
Pt = The projection onto H 2 H i1 H i2 in H.
n=0 (i1 , ,in )n

Lemma 4.2. Keep the above notations.

X
M X
N
si si = P1 + Ps and tk tk = P1 + Pt . (4.3)
i=1 k=1
14
Hence
X
M X
N
si si + tk tk + P0 = 1F (H) + P1 . (4.4)
i=1 k=1
Proof. For 1 i1 2 i2 in1 n Fn (H) with 2 n N, we have
si si (1 i1 2 i2 in1 n )
(
ui 1 1 (ui | 1 )2 i2 in1 n ) if i1 = 1,
=
0 if i1 = 2.
PM
As ui 1 1 (ui | 1 B1 )2 = ui 1 (ui | 1 B1 )1 2 and i=1 ui 1 (ui | 1 B1 ) = 1 ,
we have
(
XM
1 1 2 i2 in1 n if i1 = 1,

si si (1 i1 2 i2 in1 n ) =
i=1
0 if i1 = 2
and hence
X
M
si si |
n=2 Fn (H)
= Ps |
n=2 Fn (H)
.
i=1
For F1 (H) = H, we have si si = si (ui | B1 0) = ui 1 (ui | B1 ) so that
X
M X
M
si si = ui 1 (ui | B1 ) =
i=1 i=1
and hence
X
M
si si |F1 (H) = 1F1 (H) .
i=1
As si si (b1 b2 ) = 0 for b1 b2 B1 B2 , we have
X
M
si si |F0 (H) = 0.
i=1
Therefore we conclude that
XM X
N
si si = Ps + P1 and similarly tk tk = Pt + P1 .
i=1 k=1
As Ps + Pt + P0 + P1 = 1F (H) , one obtaines (4.4).
We set the operators
Si = Sui (= [si ]) for i = 1, . . . , M, and
Tk = Tvk (= [ti ])
for k = 1, . . . , N
PM PN
in the C -algebra OF (H) . As two operators i=1 si si and k=1 tk tk are projections
PM PN
by (4.3), so are i=1 Si Si and k=1 Tk Tk . Since P1 P0 J(H), the identity (4.4)
implies
X
M X
N
Si Si + Tk Tk = 1. (4.5)
i=1 k=1
Therefore we have
15
Theorem 4.3. Let H be a Hibert C -quad module over (A; B1 , B2 ) of nite type
with nite basis {u1 , . . . , uM } as a right B1 -module and {v1 , . . . , vN } as a right
B2 -module. Then we have
(i) The C -algebra OF (H) is generated by the operators S1 , . . . , SM , T1 , . . . , TN
and the elements 1 (z), 2 (w) for z B1 , w B2 .
(ii) They satisfy the following operator relations:
X
M X
N
Si Si + Tk Tk = 1, Sj Tl = 0, (4.6)
i=1 k=1
Si Sj = 1 (ui | uj B1 ), Tk Tl = 2 (vk | vl B2 ), (4.7)
X
M X
N
1 (z)Sj = Si 1 (ui | 1 (z)uj B1 ), 1 (z)Tl = Tk 2 (vk | 1 (z)vl B2 ),
i=1 k=1 (4.8)
X
M X
N
2 (w)Sj = Si 1 (ui | 2 (w)uj B1 ), 2 (w)Tl = Tk 2 (vk | 2 (w)vl B2 )
i=1 k=1 (4.9)
for z B1 , w B2 , i, j = 1, . . . , M, k, l = 1, . . . , N .
(iii) There exists an action of R/Z = T on OF (H) such that

1r 1r
r (Si ) = e2 Si , r (Tk ) = e2 Tk ,
r (1 (z)) = 1 (z), r (2 (w)) = 2 (w)
for r R/Z = T, i = 1, . . . , M, k = 1, . . . , N , and z B1 , w B2 .
Proof. (i) The assertion comes from the equalities (4.2).
(ii) The rst equality of (4.6) is (4.5). As the projection P0 belongs to J(H),
Lemma 4.1 ensures us the second equality of (4.6). The equalities (4.7) come from
PM
(3.7). For z B1 and j = 1, . . . , M, we have 1 (z)uj = i=1 ui 1 (ui | 1 (z)uj B1 )
so that
X
M
1 (z)sj = s1 (z)uj = sui 1 (ui | 1 (z)uj B1 )
i=1

which goes to the rst equality of (4.8). The other equalities of (4.8) and (4.9) are
similarly shown.
(iii) The assertion is direct from Lemma 3.7.

The action of T on OF (H) dened in the above theorem (iii) is called the gauge
action.

5. The universal C -algebras associated with Hilbert C -quad

modules
In this section, we will prove that the C -algebra OF (H) associated with a Hilbert
C -quad module of nite type is the universal C -algebra subject to the operator

relations stated in Theorem 4.3 (ii). Throughout this section, we x a Hilbert C -

quad module H over (A; B1 , B2 ) of nite type with nite basis {u1 , . . . , uM } as a
right Hilbert B1 -module and {v1 , . . . , vN } as a right Hilbert B2 -module as in the
previous section.
16
 Let PH be the universal -algebra generated by operators S1 , . . . , SM , T1 , . . . , TN
and elements z B1 , w B2 subject to the relations:
X
M X
N
Si Si + Tk Tk = 1, Sj Tl = 0, (5.1)
i=1 k=1
Si Sj = ui | uj B1 , Tk Tl = vk | vl B2 , (5.2)
X
M X
N
zSj = Si ui | 1 (z)uj B1 , zTl = Tk vk | 1 (z)vl B2 , (5.3)
i=1 k=1
X
M X
N
wSj = Si ui | 2 (w)uj B1 , wTl = Tk vk | 2 (w)vl B2 (5.4)
i=1 k=1

for z B1 , w B2 , i, j = 1, . . . , M, k, l = 1, . . . , N . The above four relations

(5.1),(5.2),(5.3),(5.4) are called the relations (H). In what follows, we x operators
S1 , . . . , SM , T1 , . . . , TN satisfying the relations (H).
PM PN
Lemma 5.1. The sums i=1 Si Si and k=1 Tk Tk are both projections.
PM PN
Proof. Put P = i=1 Si Si and Q = k=1 Tk Tk . By the relations (5.1), one sees
that 0 P, Q 1 and P + Q = 1, P Q = 0. It is easy to see that both P and Q are
projections.

Lemma 5.2.
(i) For i, j = 1, . . . , M and z B1 , w B2 we have
Si zSj = ui | 1 (z)uj B1 , Si wSj = ui | 2 (w)uj B1 .
(ii) For k, l = 1, . . . , N and z B1 , w B2 we have
Tk zTl = vk | 1 (z)vl B2 , Tk wTl = vk | 2 (w)vl B2 .
Proof. (i) By (5.3), we have
X
M X
M
Si zSj = Si Sh uh | 1 (z)uj B1 = ui | uh B1 uh | 1 (z)uj B1
h=1 h=1
X
M
= ui | uh uh | 1 (z)uj B1 B1 = ui | 1 (z)uj B1 .
h=1

The other equality Si wSj = ui | 2 (w)uj B1 is similarly shown to the above

equalities.
(ii) is similar to (i).

By the equalities (2.10), (2.11) we have

Lemma 5.3. Keep the above notations.
(i) For w B2 , j = 1, . . . M , the element Sj wSj belongs to A and the formula
holds:
X
M
Sj | B2 Sj = | A for , H.
j=1

17
 (ii) For z B1 , l = 1, . . . N , the element Tl zTl belongs to A and the formula
holds:
X N
Tl | B1 Tl = | A for , H.
l=1

Lemma 5.4. The following equalities for z B1 and w B2 hold:

(i)
X
M X
N
z= Si ui | 1 (z)uj B1 Sj + Tk vk | 1 (z)vl B2 Tl , (5.5)
i,j=1 k,l=1

X
M X
N
w= Si ui | 2 (w)uj B1 Sj + Tk vk | 2 (w)vl B2 Tl . (5.6)
i,j=1 k,l=1

(ii)
X
M X
N
z = Si ui | 1 (z) uj B1 Sj + Tk vk | 1 (z) vl B2 Tl , (5.7)
i,j=1 k,l=1

X
M X
N
w = Si ui | 2 (w) uj B1 Sj + Tk vk | 2 (w) vl B2 Tl . (5.8)
i,j=1 k,l=1

(iii)
X
M X
N
zw = Si ui | 1 (z)2 (w)uj B1 Sj + Tk vk | 1 (z)2 (w)vl B2 Tl ,
i,j=1 k,l=1

X
M X
N
wz = Si ui | 2 (w)1 (z)uj B1 Sj + Tk vk | 2 (w)1 (z)vl B2 Tl .
i,j=1 k,l=1

Proof. (i) By (5.3) and (5.4), we have

X
M X
N
zSj Sj = Si ui | 1 (z)uj B1 Sj , zTl Tl = Tk vk | 1 (z)vl B2 Tl
i=1 k=1
so that by (5.1)
X
M X
N
z= Si ui | 1 (z)uj B1 Sj + Tk vk | 1 (z)vl B2 Tl .
i,j=1 k,l=1

Similarly we have (5.6).

(ii) All the adjoints of 1 (z), 2 (w) for z B1 , w B2 by the three inner products
| B1 , | B2 and | A on H coincide with 1 (z ), 2 (w ) respectively. Hence
the assertions are clear.
(iii) By (i) we have
X
M X
N
zw =( Si ui | 1 (z)uj B1 Sj + Tk vk | 1 (z)vl B2 Tl )
i,j=1 k,l=1

X
M X
N
( Sg ug | 2 (w)uh B1 Sh + Tm vm | 2 (w)vn B2 Tn ).
g,h=1 m,n=1
18
As Sj Tm = Tl Sg = 0 for any j, g = 1, . . . , M, l, m = 1, . . . , N , it follows that

X
M
zw = Si ui | 1 (z)uj B1 Sj Sg ug | 2 (w)uh B1 Sh
i,j,g,h=1

X
N
+ Tk vk | 1 (z)vl B2 Tl Tm vm | 2 (w)vn B2 Tn
k,l,m,n=1

X
M
= Si ui | 1 (z)uj B1 uj | ug B1 ug | 2 (w)uh B1 Sh
i,j,g,h=1

X
N
+ Tk vk | 1 (z)vl B2 vl | vm B2 vm | 2 (w)vn B2 Tn
k,l,m,n=1

X
M X
M
= Si ui | 1 (z) uj uj | ug B1 B1 ug | 2 (w)uh B1 Sh
i,g,h=1 j=1

X
N X
N
+ Tk vk | 1 (z) vl vl | vm B2 B2 vm | 2 (w)vn B2 Tn
k,m,n=1 l=1

X
M X
M
= Si ui | 1 (z) ug B1 ug | 2 (w)uh B1 Sh
i,h=1 g=1

X
N X
N
+ Tk vk | 1 (z) vm B2 vm | 2 (w)vn B2 Tn
k,n=1 m=1

X
M X
N
= Si ui | 1 (z)2 (w)uh B1 Sh + Tk vk | 1 (z)2 (w)vn B2 Tn .
i,h=1 k,n=1

Lemma 5.5. Let p(z, w) be a polynomial of elements of B1 and B2 . Then we have

PM
(i) p(z, w)Sj = i=1 Si zi for some zi B1 .
PN
(ii) p(z, w)Tl = k=1 Tk wk for some wk B2 .
Proof. For z B1 , w B2 and i, j = 1, . . . , M , by putting zi,j = ui | 1 (z)uj B1
B1 and wi,j = ui | 2 (w)uj B1 B1 , the relations (5.3),(5.4) imply

X
M X
M
zSj = Si zi,j , wSj = Si wi,j
i=1 i=1

so that the assertion of (i) holds. (ii) is similarly shown to (i).

Lemma 5.6. Let p(z, w) be a polynomial of elements of B1 and B2 . Then we have

(i) Si p(z, w)Sj belongs to B1 for all i, j = 1, . . . , M .
(ii) Tk p(z, w)Tl belongs to B2 for all k, l = 1, . . . , N .
(iii) Si p(z, w)Tl = 0 for all i = 1, . . . , M, l = 1, . . . , N .
(iv) Tk p(z, w)Sj = 0 for all k = 1, . . . , N, j = 1, . . . , M .
19
Proof. (i) By the previous lemma, we know
X
M
p(z, w)Sj = Sh zh for some zh B1 ,
h=1

so that
X
M X
M
Si p(z, w)Sj = Si Sh zh = ui | uh B1 zh .
h=1 h=1
As ui | uh B1 zh belongs to B1 , we see the assertion. (ii) is similarly shown to (i).
(iii) As Tk Si = 0, we have
X
M
Tk p(z, w)Sj = Tk Si zi = 0.
i=1
(iv) is similarly shown to (i).
We set
S1,i := Si , i = 1, . . . , M, S2,k := Tk , k = 1, . . . , N. (5.9)
Put
1 = {(1, i) | i = 1, . . . , M }, 2 = {(2, k) | k = 1, . . . , N }.
Lemma 5.7. Every element of PH can be written as a linear combination of ele-
ments of the form
Sg1 ,i1 Sg2 ,i2 Sgm ,im bShn ,jn Sh2 ,j2 Sh1 ,j1
for some (g1 , i1 ), (g2 , i2 ), . . . , (gm , im ), (h1 , j1 ), (h2 , j2 ), . . . , (hn , jn ) 1 2 where
b is a polynomial of elements of B1 and B2 .
Proof. The assertion follows from the preceding lemmas.
By construction, every representation of B1 and B2 on a Hilbert space H together
with operators Si , i = 1, . . . , M, Tk , k = 1, . . . , N satisfying the relations (H) ex-
tends to a representation of PH on B(H). We will endow PH with the norm
obtained by taking the supremum of the norms in B(H) over all such representa-
tions. Note that this supremum is nite for every element of PH because of the
inequalities Si , Tk 1, which come from (5.1). The completion of the alge-
bra PH under the norm becomes a C -algebra denoted by OH , which is called the
universal C -algebra subject to the relations (H).
Denote by C (1 (B1 ), 2 (B2 )) the C -subalgebra of LA (H) generated by 1 (B1 )
and 2 (B2 ).
Lemma 5.8. An element L of the C -algebra C (1 (B1 ), 2 (B2 )) is both a right
B1 -module map and a right B2 -module map. This means that the equalities
[L]i (bi ) = L[i (bi )] for H, bi Bi
hold.
Proof. Since both the operators 1 (z) for z B1 and 2 (w) for w B2 are right
Bi -module maps for i = 1, 2, any element of the -algebra algebraically generated
by 1 (B1 ) and 2 (B2 ) is both a right B1 -module map and a right B2 -module map.
Hence it is easy to see that any element L of the C -algebra C (1 (B1 ), 2 (B2 )) is
both a right B1 -module map and a right B2 -module map.
20
Denote by B the C -subalgebra of OH generated by B1 and B2 .

Lemma 5.9. The correspondence

z, w B 1 (z), 2 (w) C (1 (B1 ), 2 (B2 )) LA (H) (5.10)

gives rise to an isomorphism from B onto C (1 (B1 ), 2 (B2 )) as C -algebras.

Proof. We note that by hypothesis both the maps

1 : z B1 1 (z) LA (H),
2 : w B2 2 (w) LA (H)

are injective. Denote by P (1 (B1 ), 2 (B2 )) the -algebra on H algebraically gen-

erated by 1 (z), 2 (w) for z B1 , w B2 . Dene an operator (L) OH for
L P (1 (B1 ), 2 (B2 )) by

X
M X
N
(L) = Si ui | Luj B1 Sj + Tk vk | Lvl B2 Tl . (5.11)
i,j=1 k,l=1

Let P be the -subalgebra of PH algebraically generated by B1 and B2 . Since

(1 (z)) = z for z B1 and (2 (w)) = w for w B2 and by Lemma 5.4, the map

: P (1 (B1 ), 2 (B2 )) P B

yields a -homomorphism. As Si Tk = 0 for i = 1, . . . , M, k = 1, . . . , N, we have

X
M X
N
(L) = Max{ Si ui | Luj B1 Sj , Tk vk | Lvl B2 Tl }.
i,j=1 k,l=1

We then have

X
M X
M X
M
Si ui | Luj B1 Sj ui | Luj B1 ( ui B1 uj B1 )L
i,j=1 i,j=1 i,j=1

and similarly

X
N X
N
Tk vk | Lvl B2 Tl ( vk B2 vl B2 )L.
k,l=1 k,l=1

PM PN
By putting C = Max{ i,j=1 ui B1 uj B1 , k,l=1 vk B2 vl B2 }, one has

(L) CL for all L P (1 (B1 ), 2 (B2 )).

Hence extends to the C -algebra C (1 (B1 ), 2 (B2 )) such that (C (1 (B1 ), 2 (B2 ))) =
B . The equality (5.11) holds for L C (1 (B1 ), 2 (B2 )).
21
 We will next show that : C (1 (B1 ), 2 (B2 )) B is injective. By (5.11), we
have for L C (1 (B1 ), 2 (B2 )) and h, h = 1, . . . , M ,

X
M
Sh (L)Sh = Sh Si ui | Luj B1 Sj Sh
i,j=1

X
M
= uh | ui B1 ui | Luj B1 uj | uh B1
i,j=1

X
M X
M
= uh | ui ui | Luj B1 B1 uj | uh B1
j=1 i=1

X
M
= uh | Luj B1 uj | uh B1 = uh | Luh B1 .
j=1

Suppose that (L) = 0 so that uh | Luh B1 = 0. Since

X
M
Lu h = uh uh | Luh B1 ,
h=1

we see that Luh = 0 so that L = 0. We thus conclude that is injective and hence
isomorphic.

Denote by : B C (1 (B1 ), 2 (B2 )) the inverse 1 of the -isomorphism

giving in the proof of the above lemma which satises

(z) = 1 (z) for z B1 , (w) = 2 (w) for w B2 .

We put FH 0
= B . For n N, we denote by FH
n
the closed linear span of elements
of the form

Sg1 ,i1 Sg2 ,i2 Sgn ,in bShn ,jn Sh2 ,j2 Sh1 ,j1

for some (g1 , i1 ), (g2 , i2 ), . . . , (gn , in ), (h1 , j1 ), (h2 , j2 ), . . . , (hn , jn ) 1 2 and

b B . Let us denote by FH the C -subalgebra of OH generated by n=0 FH . By
n

the relations (5.5) and (5.6), we see the following

Lemma 5.10. For x B , the following identity holds:

X
M X
N
x= Si ui | (x)uj B1 Sj + Tk vk | (x)vl B2 Tl . (5.12)
i,j=1 k,l=1

Hence by putting for b B

b1,ij = ui | (b)uj B1 , i, j = 1, . . . , M,
b2,kl = vk | (b)vl B2 , k, l = 1, . . . , N,

we have
22
Lemma 5.11. For b B , the identity

Sg1 ,i1 Sg2 ,i2 Sgn ,in bShn ,jn Sh2 ,j2 Sh1 ,j1
X
M

= Sg1 ,i1 Sg2 ,i2 Sgn ,in S1,i b1,ij S1,j Shn ,jn Sh2 ,j2 Sh1 ,j1
i,j=1

X
N

+ Sg1 ,i1 Sg2 ,i2 Sgn ,in S2,k b2,kl S2,l Shn ,jn Sh2 ,j2 Sh1 ,j1
k,l=1

holds and induces an embedding of FH

n
, FH
n+1
for n Z+ .

Lemma 5.12. The C -algebra FH is the inductive limit limn FH

n
of the sequence
of the inclusions:

FH
0
, FH
1
, FH
2
, , FH
n
, FH
n+1
, , FH . (5.13)

1r
Let e2 T be a complex number of modulus one for r R/Z. The elements

1r 1r
e2 Si , i = 1, . . . , M, e2 Tk , k = 1, . . . , N, z B1 , w B2

in OH instead of

Si , i = 1, . . . , M, Tk , k = 1, . . . , N, z B1 , w B2

satisfy the relations (H). This implies the existence of an action on PH by auto-
morphisms of the one-dimensional torus T that acts on the generators by

1r 1r
hr (Si ) = e2 Si , hr (Tk ) = e2 Tk ,
hr (z) = z, hr (w) = w

for i = 1, . . . , M, k = 1, . . . , N , z B1 , w B2 and r R/Z = T. As the C -algebra

OH has the largest norm on PH , the action (hr )rT on PH extends to an action of
T on OH , still denote by h. The formula
Z
a OH hr (a)dr OH
rT

where dr is the normalized Lebesgue measure on T denes a faithful conditional

expectation denoted by EH from OH onto the xed point algebra (OH )h . The
following lemma is routine.

Lemma 5.13. (OH )h = FH .

The C -algebra OH satises the following universal property. Let D be a unital

C -algebra and 1 : B1 D, 2 : B2 D be -homomorphisms such that
1 (a) = 2 (a) for a A. Assume that there exist elements Sb1 , . . . , SbM , Tb1 , . . . , TbN
23
in D satisfying the relations:
X
M X
N
Sbi Sbi + Tbk Tbk = 1, Sbj Tbl = 0,
i=1 k=1

Sbi Sbj = 1 (ui | uj B1 ), Tbk Tbl = 2 (vk | vl B2 ),

X
M X
N
1 (z)Sbj = Sbi 1 (ui | 1 (z)uj B1 ), 1 (z)Tbl = Tbk 2 (vk | 1 (z)vl B2 ),
i=1 k=1
X
M X
N
2 (w)Sbj = Sbi 1 (ui | 2 (w)uj B1 ), 2 (w)Tbl = Tbk 2 (vk | 2 (w)vl B2 ),
i=1 k=1

for z B1 , w B2 , i, j = 1, . . . , M, k, l = 1, . . . , N , then there exists a unique

-homomorphism : OH D such that

(Si ) = Sbi , (Tk ) = Tbk , (z) = 1 (z), (w) = 2 (w)

for i = 1, . . . , M, k = 1, . . . , N and z B1 , w B2 . We further assume that both the
homomorphisms i : Bi D, i = 1, 2 are injective. We denote by : B D
the restriction of to the subalgebra B . Let us denote by O bH the C -subalgebra
of D generated by Sbi , Tbk , i = 1, . . . , M, k = 1, . . . , N and 1 (z), 2 (w) for z B1 ,
w B2 .
Lemma 5.14. Keep the above situation. The -homomorphism : B D is
injective.
Proof. Since the correspondence in Lemma 5.9
: z, w, B 1 (z), 2 (w) C (1 (B1 ), 2 (B2 ))
yields an isomorphism of C -algebras, it suces to prove that the correspondence
1 (z), 2 (w) C (1 (B1 ), 2 (B2 )) 1 (z), 2 (w) D

yields an isomorphism. Let Bb be the C -subalgebra of O

bH generated by elements
1 (z), 2 (w) A for z B1 , w B2 . Dene an element b(L) of D for L
C (1 (B1 ), 2 (B2 )) by setting
X
M X
N
b(L) =
Sbi 1 (ui | Luj B1 )Sbj + Tbk 2 (vk | Lvl B2 )Tbl D.
i,j=1 k,l=1 (5.14)
b gives rise to a -homomorphism from
As in the proof of Lemma 5.9, one sees that
C (1 (B1 ), 2 (B2 )) into D. Since
X
M X
N
b(1 (z)) =
Sbi 1 (ui | 1 (z)uj B1 )Sbj + Tbk 2 (vk | 1 (z)vl B2 )Tbl = 1 (z),
i,j=1 k,l=1

and similarly b(2 (w)) = 2 (w), it is enough to show that b is injective. Suppose
that b(L) = 0 for some L C (1 (B1 ), 2 (B2 )). By following the proof of Lemma
b(L)Sbh = 1 (uh | Luh B1 ) for all h, h = 1, . . . , M . Hence
5.9, one sees that Sbh
the condition b(L) = 0 implies 1 (uh | Luh B1 ) = 0. Since 1 is injective, we
24
have uh | Luh B1 = 0 for all h, h = 1, . . . , M . As L is a right B1 -module map, we
have for H,
X
M X
M
L = L(uh uh | B1 ) = (Luh )uh | B1 = 0
h =1 h =1

so that L = 0. Therefore b : C (1 (B1 ), 2 (B2 )) D is injective. Hence the

composition
b
b : B C (1 (B1 ), 2 (B2 )) D

:
is injective.

We set
Sb1,i := Sbi , i = 1, . . . , M, Sb2,k := Tbk , k = 1, . . . , N. (5.15)

We put FbH
0
= B . For n N, let FbH
n
be the closed linear span in the C -algebra
bH of elements of the form
O
Sbg1 ,i1 Sbg2 ,i2 Sbgn ,in (b)Sbhn ,jn Sbh2 ,j2 Sbh1 ,j1
for (g1 , i1 ), (g2 , i2 ), . . . , (gn , in ), (h1 , j1 ), (h2 , j2 ), . . . , (hn , jn ) 1 2 and b B .
Similarly to the subalgebras FH n
, n Z+ , of OH , one knows that the closed linear
span FH is a C -algebra and naturally regarded as a subalgebra of FbH
b n n+1
for each
n Z+ . Let us denote by FbH the C -subalgebra of O bH generated by Fbn .
n=0 H
Then the C -algebra FbH is the inductive limit limn FbH n
of the sequence of the
inclusions :
FbH
0
, FbH
1
, FbH
2
, , FbH
n
, FbH
n+1
, , FbH . (5.16)

Lemma 5.15. Suppose that both the -homomorphisms i : Bi O bH , i = 1, 2

are injective. Then the restriction of to the subalgebra FH yields a -isomorphism
|FH : FH FbH .
Proof. By the universality of OH , the restriction of to FH yields a surjective
-homomorphism |FH : FH FbH . It suces to show that |FH is injective.
Suppose that Ker(|FH ) = {0} and put I = Ker(|FH ). Since F (FHn
) = FbH
n
and
FH = limn FH , there exists n Z+ such that I FH = 0. Let us denote by n
n n

the set of n-tuples of 1 2 :

n = {(1 , . . . , n ) | 1 , . . . , n 1 2 }.
For = (1 , . . . , n ) n , denote by S the operator
S = S1 S2 Sn
where (
S1,i = Si if m = (1, i) 1 ,
Sm =
S2,k = Tk if m = (2, k) 2 , m = 1, . . . , n.
Any element of FH
n
is of the form
X
S b, S for some b, B .
,n

25
 P PM
Hence one may nd a nonzero element ,n S b, S IFH n
. Since i=1 Si Si +
PN
P
n S S = 1 hols. For any , , one then
n
k=1 Tk Tk = 1, the equality
sees
X
0 = S ( S b, S )S I FH
n
.
,n

As Si Tk = 0 and Si Sj = u
Pi | uj B1 , Tk Tl = vk | vl B2 for i, j = 1, . . . , M, k, l =
1, . . . , N , the element S ( ,n S b, S )S belongs to I B . By the pre-
ceding P lemma, the homomorphism : B O bH is injective, so that we have
(S ( ,n S b, S )S ) = 0 a contradiction. Therefore we conclude that
|F : FH FbH is injective and hence isomorphic.
H

The following theorem is one of the main results of the paper.

Theorem 5.16. Let D be a unital C -algebra. Suppose that there exist -homomorphisms
1 : B1 D, 2 : B2 D such that 1 (a) = 2 (a) for a A and there exist
elements Sb1 , . . . , SbM , Tb1 , . . . , TbN in D satisfying the relations:
X
M X
N
Sbi Sbi + Tbk Tbk = 1, Sj Tl = 0,
i=1 k=1

Sbi Sbj = 1 (ui | uj B1 ), Tbk Tbl = 2 (vk | vl B2 ),

X
M X
N
1 (z)Sbj = Sbi 1 (ui | 1 (z)uj B1 ), 1 (z)Tbl = Tbk 2 (vk | 1 (z)vl B2 ),
i=1 k=1
X
M X
N
2 (w)Sbj = Sbi 1 (ui | 2 (w)uj B1 ), 2 (w)Tbl = Tbk 2 (vk | 2 (w)vl B2 ),
i=1 k=1

bH the C -
for z B1 , w B2 , i, j = 1, . . . , M, k, l = 1, . . . , N . Let us denote by O
b b
subalgebra of D generated by Si , Tk , i = 1, . . . , M, k = 1, . . . , N and 1 (z), 2 (w),
for z B1 , w B2 . We further assume that the algebra O bH admits a gauge action.
If both the -homomorphisms i : Bi A, i = 1, 2 are injective, then there exists
a -isomomorphism : OH O bH satisfying

(Si ) = Sbi , (Tk ) = Tbk , (z) = 1 (z), (w) = 2 (w)

(5.17)
for i = 1, . . . , M, k = 1, . . . , N and z B1 , w B2 .

Proof. By assumption, O bH admits a gauge action, which we dnote by b h. Let us

b b
h b b
denote by (OH ) the xed point algebra of OH under the gauge action h and by FbH
the C -subalgebra of O bH dened by the inductive limit (5.16). Then it is routine
to check that (ObH )bh is canonically -isomorphic to FbH . There exists a conditional
expectation
EbH : O
bH FbH
dened by
Z
EbH (x) = b
hr (x)dr bH .
for x O
rT
26
By the universality of the algebra OH there exists a surjective -homomorphism
from OH to ObH such that

(Si ) = Sbi , (Tk ) = Tbk , (z) = 1 (z), (w) = 2 (w)

for i, j = 1, . . . , M, k, l = 1, . . . , N, z B1 , w B2 . Then (FH ) = FbH and the
following diagram:

OH O bH


EH y

yb
EH

|F
FH FbH
H

is commutative. Denote by the restriction of to the C -subalgebra B of OH

generated by z B1 , w B2 . By assumption, both the maps i : Bi O bH , i =
1, 2 are injective, so that : B ObH is injective by Lemma 5.14. By the
preceding lemma, |FH : FH FbH is an isomorphism. Since the conditional
expectation EH : OH FH is faithful, a routine argument shows that is
injective and hence isomorphic.
Therefore we have
Theorem 5.17. For a C -quad module H of nite type, the C -algebra OF (H)
generated by the quotients [s ], [t ] of the creation operators s , t for H on
the Fock spaces F (H) is canonically isomorphic to the universal C -algebra OH
generated by operators S1 , . . . , SM , T1 , . . . , TN and elements z B1 , w B2 subject
to the relations:
X
M X
N
Si Si + Tk Tk = 1, Sj Tl = 0,
i=1 k=1
Si Sj = ui | uj B1 , Tk Tl = vk | vl B2 ,
X
M X
N
zSj = Si ui | 1 (z)uj B1 , zTl = Tk vk | 1 (z)vl B2 ,
i=1 k=1
X
M X
N
wSj = Si ui | 2 (w)uj B1 , wTl = Tk vk | 2 (w)vl B2 ,
i=1 k=1

for i, j = 1, . . . , M, k, l = 1, . . . , N and z B1 , w B2 .
Proof. Theorem 4.3 implies that the operators S1 , . . . , SM , T1 , . . . , TN and the el-
ements 1 (z), 2 (w) for z B1 , w B2 in OF (H) satisfy the eight relations of
Theorem 5.16. By Theorem 5.16, we see that the correspondences
Si Si , Tk Tk , z B1 1 (z), w B2 2 (w)
for i = 1, . . . , M, k = 1, . . . , N, and z B1 , w B2 give rise to an isomorphism
from OH to OF (H) .
The eight relations of the operators above are called the relations (H). The above
generating operators S1 , . . . , SM and T1 , . . . , TN of the universal C -algebra OH
correspond to two nite basis {u1 , . . . , uM } and {v1 , . . . , vN } of the Hilbert C -
quad module H respectively. On the other hand, the other C -algebra OF (H) is
generated by the quotients of the creation operators s , t for H on the Fock
27
spaces F (H), which do not depend on the choice of the two nite bases. Hence we
have
Corollary 5.18. For a C -quad module H of nite type, the universal C -algebra
OH generated by operators S1 , . . . , SM , T1 , . . . , TN and elements z B1 , w B2 sub-
ject to the relations (H) does not depend on the choice of the nite bases {u1 , . . . , uM }
and {v1 , . . . , vN }.

6. K-Theory formulae
Let H be a Hilbert C -quad module over (A; B1 , B2 ) of nite type as in the
preceding section. In this section, we will state K-theory formulae for the C -
algebra OF (H) . By the previous section, the C -algebra OF (H) is regarded as the
universal C -algebra OH generated by the operators S1 , . . . , SM and T1 , . . . , TN
and the elements z B1 and w B2 subject to the relations (H). Let us denote by
B the C -subalgebra of OH generated by elements z B1 and w B2 . By Lemma
5.9 the correspondence
z, w B 1 (z), 2 (w) C (1 (B1 ), 2 (B2 )) LA (H) (6.1)

gives rise to a -isomorphism from B onto C (1 (B1 ), 2 (B2 )) as C -algebras,

which is denoted by . We will restrict our interest to the case when
(i) S1 , . . . , SM and T1 , . . . , TN are partial isometries, and

(ii) S1 S1 , . . . , SM SM , T1 T1 , . . . , TN TN commute with all elements of B .
If the bases {u1 , . . . , uM } and {v1 , . . . , vN } satisfy the conditions
ui | uj B1 = 0 for i = j, vk | vl B2 = 0 for k = l,

the condition (i) holds. Furthermore if 1 (z) acts diagonally on {u1 , . . . , uM } for
z B1 and 2 (w) acts diagonally on {v1 , . . . , vN } for w B2 , th condition (ii)
holds. Rcall that the gauge action is denoted by h which is an action of T on OH
such that the xed point algebra (OH )h under h is canonically isomorphic to the
C -algebra FH . Denote by b
h the dual action of h which is an action of Z = T b on the
C -crossed product OH h T by the gauge action h of T. As in the argument of [17],
OH h T is stably isomorphic to FH . Hence we have K (OH h T) is isomorphic to
K (FH ). The dual action bh induces an automorphism on the group K (OH h T)
and hence on K (FH ), which is denoted by . Then by [17] (cf. [3], [21], etc.) we
have
Proposition 6.1. The following six term exact sequence of K-theory hold:
id

K0 (FH ) K0 (FH ) K0 (OH )
x

y
K1 (OH ) K1 (FH ) K1 (FH ).
id

We put for x B
1,i (x) = Si xSi , i = 1, . . . , M,
2,k (x) = Tk xTk , k = 1, . . . , N.
28
Both the families 1,i , 2,k yield endomorphisms on B which give rise to endomor-
phisms on the K-groups:
1,i :K0 (B ) K0 (B ), i = 1, . . . , M,
2,k :K0 (B ) K0 (B ), k = 1, . . . , N.
PM PN
We put = i=1 1,i + k=1 2,k which is an endomorphism on K0 (B ). Now
we further assume that K1 (FH ) = {0}. It is routine to show that the groups
Coker(id ) in K0 (FH ) and Ker(id ) in K0 (FH ) are isomorphic to the groups
Coker(id ) in K0 (B ) and Ker(id ) in K0 (B ) respectively by an argument
of [3]. Therefore we have
Proposition 6.2.
K0 (OH ) = Coker(id ) in K0 (B ),
K1 (OH ) = Ker(id ) in K0 (B ).

7. Examples
In this section, we will study the C -algebras OH for the Hilbert C -quad mod-
ules presented in Examples in Section 2.
1. Let , be automorphisms of a unital C -algebra A satisfying = .
Let H, be the associated Hilbert C -quad module of nite type as in 1 in Section
2. It is easy to see the following proposition.
Proposition 7.1. The C -algebra OH, associated to the Hilbert C -quad module
H, coming from commuting automorphisms , of a unital C -algebra A is iso-
morphic to the universal C -algebra generated by two isometries U, V and elements
x of A subject to the following relations:
U U + V V = 1,
U U x = xU U , V V x = xV V ,
(x) = U xU, (x) = V xV
for x A.

2. We x natural numbers 1 < N, M N. Consider nite dimensional com-

mutative C -algebras A = C, B1 = CN , B2 = CM . The algebras B1 , B2 have the
ordinary product structure and the inner product structure which we denote by
| N and | M respectively. Let us denote by HM,N the Hilbert C -quad
module CM CN over (C; CN , CM ) dsened in 2 in Section 2. Put the nite bases
ui = ei 1 HM,N , i = 1, . . . , M as a right B1 module, and
vk = 1 fk HM,N , k = 1, . . . , N as a right B2 module.
We set = {(i, k) | 1 i M, 1 k N } and put e(i,k) = ei fk , (i, k)
the standard basis of HM,N . Then the C -algebra B on HM,N generated by B1
and B2 is regarded as CM CN = B2 B1 . Hence
X
B = Ce(i,k) .
(i,k)

29
Lemma 7.2. The C -algebra OHM,N is generated by operators Si , Tk , e(i,k) , i =
1, . . . , M, k = 1, . . . , N satisfying
X
M X
N
Si Si + Tk Tk = 1, (7.1)
i=1 k=1
Si Sj = i,j , Tk Tl = k,l , (7.2)
X
M X
N
e(i,k) Sj = i,j Sj e(h,k) , e(i,k) Tl = k,l Tl e(i,m) (7.3)
h=1 m=1

for i, j = 1, . . . , M, k, l = 1, . . . , N.
Proof. It suces to show the equalities (7.3). We have
e(i,k) Sj = Sj uj | (e(i,k) )uj B1
= Sj ej 1 | (ei fk )(ej 1)B1
X
M
= i,j Sj (1 fk ) = i,j Sj e(h,k) .
h=1

The other equality of (7.3) is similarly shown.

Put
S(i,k) = e(i,k) Si , T(i,k) = e(i,k) Tk for (i, k) .
Then we have
Lemma 7.3.

e(i,k) = S(i,k) S(i,k) + T(i,k) T(i,k) , (7.4)
X
N X
M
Si = S(i,k) , Tk = T(i,k) , (7.5)
k=1 i=1
X X

S(i,k) S(i,k) + T(i,k) T(i,k) = 1, (7.6)
(i,k) (i,k)

X
M

S(i,k) S(i,k) = (S(j,k) S(j,k) + T(j,k) T(j,k) ), (7.7)
j=1

X
N

T(i,k) T(i,k) = (S(i,l) S(i,l) + T(i,l) T(i,l) ) (7.8)
l=1

for i = 1, . . . , M, k = 1, . . . , N and (i, k) .

Proof. Since e(i,k) Sj = i,j e(i,k) Si , we have
XM

S(i,k) S(i,k) = e(i,k) Si Si e(i,k) = e(i,k) ( Sj Sj )e(i,k)
j=1

PN
and similarly T(i,k) T(i,k) = e(i,k) ( l=1 Tl Tl )e(i,k) . Hence we have
XM X
N
e(i,k) = e(i,k) ( Sj Sj + Tl Tl )e(i,k) = S(i,k) S(i,k)

+ T(i,k) T(i,k) ,
j=1 l=1
30
 P
so that (7.4) holds. As 1 = (j,k) e(j,k) , the equality (7.6) holds. Since e(j,k) Si =
0 for j = i, we have
X X
N
Si = ( e(j,k) )Si = S(i,k)
(j,k) k=1
PM
and similarly Tk = i=1 T(i,k) , so that (7.5) holds. By (7.3), it follows that
X
M X
M X
M

S(i,k) S(i,k) = Si Si e(j,k) = e(j,k) =
(S(j,k) S(j,k)
+ T(j,k) T(j,k) ),
j=1 j=1 j=1

and smilarly we have

X
N

T(i,k) T(i,k) = (S(i,l) S(i,l) + T(i,l) T(i,l) ).
l=1

Theorem 7.4. The C -algebra OHM,N associated with the Hilbert C -quad module
HM,N = CM CN is generated by partial isometries S(i,k) , T(i,k) for (i, k) =
{(i, k) | i = 1, . . . , M, k = 1, . . . , N } satisfying the relations:
X X

S(i,k) S(i,k) + T(i,k) T(i,k) = 1,
(i,k) (i,k)

X
M

S(i,k) S(i,k) = S(j,k) S(j,k) + T(j,k) T(j,k ),
j=1

X
N

T(i,k) T(i,k) = (S(i,l) S(i,l) + T(i,l) T(i,l) )
l=1

for (i, k) .
Proof. By the preceding lemma, one knows that e(i,k) , Si , Tk are generated by the
operators S(i,k) , T(i,k) so that the algebra OHM,N is generated by the partial isome-
tries S(i,k) , T(i,k) , (i, k) .

Let In be the n n identity matrix and En the n n matrix whose entries are all
1 s. For an M M -matrix C = [ci,j ]M
i,j=1 and an N N -matrix D = [dk,l ]k,l=1 ,
N

denote by C D the M N M N matrix

c11 D c12 D . . . c1M D
c21 D c22 D . . . c2M D

C D = . .. .. ..
.. . . .
cM 1 D cM 2 D ... cM M D
so that

IN IN ... IN EN 0 ... 0
IN ..
IN ... IN 0 EN
..
. .
EM IN = . .. .. .. , IM EN =

.

.. .
.. .. ..
. . . . . 0
IN IN ... IN 0 ... 0 EN
31
The index set {(i, k) | i = 1, . . . , M, k = 1, . . . , N } of the standard basis of CM CN
is ordered lexicographically from left as in the following way:
(1, 1), . . . , (1, N ), (2, 1), . . . , (2, N ), . . . , (M, 1), . . . , (M, N ).
Put the M N M N matrices
AM,N = EM IN , BM,N = IM EN
and the 2M N 2M N matrix

AM,N AM,N
HM,N = .
BM,N BM,N
Then we have
Theorem 7.5. The C -algebra OHM,N is isomorphic to the Cuntz-Krieger algebra
OHM,N for the matrix HM,N . The algebra OHM,N is simple and purely innite and is
isomorphic to the Cuntz-Krieger algebra OAM,N +BM,N for the matrix AM,N +BM,N .
Proof. By the preceding proposition, the C -algebra OHM,N is isomorphic to the
Cuntz-Krieger algebra OHM,N for the matrix HM,N . Since the matrix HM,N is
aperiodic, the algebra is simple and purely innite. The n-th column of the matrix
HM,N coincides with the n + N -th column for every n = 1, . . . , M . One sees that
the matrix AM,N + BM,N is obtained from HM,N by amalgamating them. The
precedure is called the column amalgamation and induces an isomorphism on their
Cuntz-Krieger algebras (see [15]).
In [15], the abelian groups
ZM N /(AM,N + BM,N IM N )ZM N , Ker(AM,N + BM,N IM N ) in ZM N
have been computed by using Euclidean algorithms. For the case M = 2, they are
Z/(N 2 1)Z, {0}
respectively, so that we see
K0 (OH2,N ) = Z/(N 2 1)Z, K1 (OH2,N ) = 0
(see [15] for details).
3. For a C -textile dynamical system (A, , , , , ), let H, be the C -
quad module over (A; B1 , B2 ) as in 3 in Section 2. The C -algebra OH,
has been
studied in [14].

8. Higher dimensional analogue

In this nal section, we will state a generalization of Hilbert C -quad modules
to Hilbert modules with multi actions of C -algebras.
Let A be a unital C -algebra and B1 , . . . , Bn be n-family of unital C -algebras.
Suppose that there exists a unital embedding
i : A , Bi
for each i = 1, . . . , n. Suppose that there exists a right action i of A on Bi such
that
bi i (a) Bi for bi Bi , a A, i = 1, . . . , n.
Hence Bi is a right A-module through i for i = 1, . . . , n. Let H be a Hilbert C -
bimodule over A with a right action of A, an A-valued right inner product | A
32
and a -homomorphism A from A to LA (H). It is called a Hilbert C -multi module
over (A; Bi , i = 1, . . . , n) if H has a multi structure of Hilbert C -bimodules over
Bi for i = 1, . . . , n such that for each i = 1, . . . , n there exist a right action i of Bi
on H and a left action i of Bi on H and a Bi -valued right inner product | Bi
such that i (zi ) LA (H) and
[i (zi )]j (wj ) = i (zi )[j (wj )], j (zj j (a)) = [j (zj )]a
for H, zi Bi , wj Bj , a A, i, j = 1, . . . , n and
A (a) = i (i (a)), a A, i = 1, . . . , n.
The operator i (zi ) on H is adjointable with respect to the inner product | Bi
whose adjoint i (zi ) coincides with the adjoint of i (zi ) with respect to the inner
product | A so that i (zi ) = i (zi ). We assume that the left actions i of Bi
on H for i = 1, 2 are faithful. We require the following compatibility conditions
between the right A-module structure of H and the right A-module structure of Bi
through i :
| aBi = | Bi i (a), , H, a A, i = 1, . . . , n.
We further assume that H is a full Hilbert C -bimodule with respect to the inner
product | A , | Bi for each. A Hilbert C -multi module H over (A; Bi , i =
1, . . . , n) is said to be of general type if there exists a faithful completely positive
map i : Bi A for i = 1, . . . , n such that
i (bi i (a)) = i (bi )a, bi Bi , a A,
i ( | Bi ) = | A , , H, i = 1, . . . , n.
A Hilbert C -multi module H over (A; Bi , i = 1, . . . , n) is said to be of nite type
(i) (i)
if there exists a family {u1 , . . . , uM (i) }, i = 1, . . . , n of nite bases of H as a right
Hilbert Bi -module for each i = 1, . . . , n such that
X
(i)
M
(i) (i)
uj i (uj | Bi ) = , H, i = 1, . . . , n
j=1

and
(i) (i)
uj |k (wk )uh Bi A, wk Bk , j, h = 1, . . . , M (i) ,
X
(i)
M
(i) (i)
uj | k ( | Bk )uj Bi = | A
j=1

for all , H, i, k = 1, . . . , n with i = k.

By a generalizing argument to the preceding sections, we may construct a C -
algebra OF (H) associated with the Hilbert C -multi module H by a similar manner
to the preceding sections, that is, the C -algebra generated by n-kinds of creation
(i)
operators s , H, i = 1, . . . , n on the generalized Fock space F (H) by the ideal
generated by the nite rank operators. One may show the following generalization:
Proposition 8.1. Let H be a Hilbert C -multi module over (A; Bi , i = 1, . . . , n) of
nite type with a nite basis {u1 , . . . , uM (i) } of H as a Hilbert C -right module over
(i) (i)

Bi for each i = 1, . . . , n. Then the C -algebra OF (H) generated by the n-kinds of
creation operators on the generalized Fock spaces F (H) is canonically isomorphic to
33
the universal C -algebra OH generated by the operators S1 , . . . , SM (i) and elements
(i) (i)

zi Bi for i = 1, . . . , n subject to the relations:

X X
(i)
n M
(i) (i) (i)
Sk Sk = 1, Sk (j)
Sm = 0, i = j,
i=1 k=1

X
(i)
M
(i) (i) (i) (i) (i) (i) (i) (i)
Sk Sl = uk | ul Bi , zj Sk = Sl ul | j (zj )uk Bi
l=1

for zj Bj , i, j = 1, . . . , n, k, l = 1, . . . , M (i) , m = 1, . . . , M (j) .

The proof of the above proposition is similar to the proof of Theorem 1.1.

References
[1] B. Blackadar, K-theory for Operator Algebras, Springer-Verlag, 1986.
[2] J. Cuntz, Simple C -algebras generated by isometries, Comm. Math. Phys. 57(1977), pp.
173185.
[3] J. Cuntz, A class of C -algebras and topological Markov chains II: reducible chains and the
Ext- functor for C -algebras, Inventions Math. 63(1980), pp. 2540.
[4] J. Cuntz, K-theory for certain C -algebras. II, J. Operator Theory 5(1981), pp. 101108.
[5] J. Cuntz and W. Krieger, A class of C -algebras and topological Markov chains, Invent.
Math. 56(1980), pp. 251268.
[6] V. Deaconu, C -algebras and Fell bundles associated to a textile system, preprint,
arXiv:1001.0037.
[7] R. Excel, D. Gonc alves and C. Starling, The tiling C -algebra viewed as a tight inverse
semigroup algebra, preprint, arXiv:1106.4535v1.
[8] T. Kajiwara, C. Pinzari and Y. Watatani, Ideal structure and simplicity of the C -algebras
generated by Hilbert modules, J. Funct. Anal. 159(1998), pp. 295322.
[9] T. Kajiwara and Y. Watatani, Jones index theory by Hilbert C -bimodules and K-theory,
Trans. Amer. Math. Soc., 352(2000), pp. 34293472.
[10] Y. Katayama, Generalized Cuntz algebras ON M , RIMS kokyuroku, 858(1994) pp. 8790.

[11] A. Kumjian and D. Pask, Higher rank graph C -algebras, New York J. Math. 6(2000), pp.
1-20.
[12] K. Matsumoto, Actions of symbolic dynamical systems on C -algebras, J. Reine Angew.
Math. 605(2007), pp. 23-49.
[13] K. Matsumoto, C -algebras associated with textile dynamical systems, preprint,
arXiv:1106.5092v1.
[14] K. Matsumoto, C -algebras associated with Hilbert C -quad modules of C -textile dynam-
ical systems, preprint, arXiv:1111.3091v1.
[15] K. Matsumoto, Cuntz-Krieger algebras associated with Hilbert C -quad modules of com-
muting matrices, preprint, arXiv:1201.1056v1.
[16] P. S. Muhly and B. Solel, On the simplicity of some Cuntz-Krieger algebras, Math. Scand.
83(1996), pp. 5373.
[17] W. L. Paschke, K-theory for actions of the circle group on C -algebras, J. Operator Theory
6(1981), pp. 125133.
[18] D. Pask, I. Raeburn and N. A. Weaver, A family of 2-graphs arising from two-dimensional
subshifts, Ergodic Theory Dynam. Sytems 29(2009), pp. 16131639.
[19] D. Pask, I. Raeburn and N. A. Weaver, Periodic 2-graphs arising from subshifts, preprint
arXiv:0911.0730.
[20] A. L. T. Paterson, Groupoids, inverse semigroups, and their operator algebras, Progress in
Mathematics 170, Birkh auser Boston, Basel, Berlin, 1998
[21] M. V. Pimsner, A class of C -algebras generalizing both CuntzKrieger algebras and crossed
products by Z, in Free Probability Theory, Fields Institute Communications 12(1996), pp.
189212.
[22] G. A. Pino, J. Clark, A. A. Huff and I. Raeburn, Kumjian-Pask algebras of higher rank
graphs, preprint, arXiv:1106.4361v1.
34
[23] G. Robertson and Steger, Ane buildings, tiling systems and higher rank CuntzKrieger
algebras, J. Reine Angew. Math. 513(1999), pp. 115-144.
[24] M. Rrdam, Classication of purely innite simple C -algebras I, J. Func. Anal. 131(1995),
pp. 415458.
[25] K. Schmidt, Dynamical systems of algebraic origin, Birkhauser, Basel, Boston, Berlin (1995).
[26] Y. Watatani, Index for C -subalgebras, Mem. Amer. Math. Soc. 424(1990).

Department of Mathematics, Joetsu University of Education, Joetsu 943-8512, Japan

E-mail address: [email protected]

35