A Note on Gluing Dirac Type Operators on a Mirror Quantum Two-Sphere
The goal of this paper is to introduce a class of operators, which we call quantum Dirac type operators on a noncommutative sphere, by a gluing construction from copies of noncommutative disks, subject to an appropriate local boundary condition. We show that the resulting operators have compact reso...
Gespeichert in:
| Datum: | 2014 |
|---|---|
| Hauptverfasser: | , |
| Format: | Artikel |
| Sprache: | English |
| Veröffentlicht: |
Інститут математики НАН України
2014
|
| Schriftenreihe: | Symmetry, Integrability and Geometry: Methods and Applications |
| Online Zugang: | http://dspace.nbuv.gov.ua/handle/123456789/146813 |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| Назва журналу: | Digital Library of Periodicals of National Academy of Sciences of Ukraine |
| Zitieren: | A Note on Gluing Dirac Type Operators on a Mirror Quantum Two-Sphere / S. Klimek, M. McBride // Symmetry, Integrability and Geometry: Methods and Applications. — 2014. — Т. 10. — Бібліогр.: 14 назв. — англ. |
Institution
Digital Library of Periodicals of National Academy of Sciences of Ukraine| id |
irk-123456789-146813 |
|---|---|
| record_format |
dspace |
| spelling |
irk-123456789-1468132019-02-12T01:23:22Z A Note on Gluing Dirac Type Operators on a Mirror Quantum Two-Sphere Klimek, S. McBride, M. The goal of this paper is to introduce a class of operators, which we call quantum Dirac type operators on a noncommutative sphere, by a gluing construction from copies of noncommutative disks, subject to an appropriate local boundary condition. We show that the resulting operators have compact resolvents, and so they are elliptic operators. 2014 Article A Note on Gluing Dirac Type Operators on a Mirror Quantum Two-Sphere / S. Klimek, M. McBride // Symmetry, Integrability and Geometry: Methods and Applications. — 2014. — Т. 10. — Бібліогр.: 14 назв. — англ. 1815-0659 2010 Mathematics Subject Classification: 46L99; 47B99; 81R60 DOI:10.3842/SIGMA.2014.036 http://dspace.nbuv.gov.ua/handle/123456789/146813 en Symmetry, Integrability and Geometry: Methods and Applications Інститут математики НАН України |
| institution |
Digital Library of Periodicals of National Academy of Sciences of Ukraine |
| collection |
DSpace DC |
| language |
English |
| description |
The goal of this paper is to introduce a class of operators, which we call quantum Dirac type operators on a noncommutative sphere, by a gluing construction from copies of noncommutative disks, subject to an appropriate local boundary condition. We show that the resulting operators have compact resolvents, and so they are elliptic operators. |
| format |
Article |
| author |
Klimek, S. McBride, M. |
| spellingShingle |
Klimek, S. McBride, M. A Note on Gluing Dirac Type Operators on a Mirror Quantum Two-Sphere Symmetry, Integrability and Geometry: Methods and Applications |
| author_facet |
Klimek, S. McBride, M. |
| author_sort |
Klimek, S. |
| title |
A Note on Gluing Dirac Type Operators on a Mirror Quantum Two-Sphere |
| title_short |
A Note on Gluing Dirac Type Operators on a Mirror Quantum Two-Sphere |
| title_full |
A Note on Gluing Dirac Type Operators on a Mirror Quantum Two-Sphere |
| title_fullStr |
A Note on Gluing Dirac Type Operators on a Mirror Quantum Two-Sphere |
| title_full_unstemmed |
A Note on Gluing Dirac Type Operators on a Mirror Quantum Two-Sphere |
| title_sort |
note on gluing dirac type operators on a mirror quantum two-sphere |
| publisher |
Інститут математики НАН України |
| publishDate |
2014 |
| url |
http://dspace.nbuv.gov.ua/handle/123456789/146813 |
| citation_txt |
A Note on Gluing Dirac Type Operators on a Mirror Quantum Two-Sphere / S. Klimek, M. McBride // Symmetry, Integrability and Geometry: Methods and Applications. — 2014. — Т. 10. — Бібліогр.: 14 назв. — англ. |
| series |
Symmetry, Integrability and Geometry: Methods and Applications |
| work_keys_str_mv |
AT klimeks anoteongluingdiractypeoperatorsonamirrorquantumtwosphere AT mcbridem anoteongluingdiractypeoperatorsonamirrorquantumtwosphere AT klimeks noteongluingdiractypeoperatorsonamirrorquantumtwosphere AT mcbridem noteongluingdiractypeoperatorsonamirrorquantumtwosphere |
| first_indexed |
2025-07-11T00:40:03Z |
| last_indexed |
2025-07-11T00:40:03Z |
| _version_ |
1837309013127069696 |
| fulltext |
Symmetry, Integrability and Geometry: Methods and Applications SIGMA 10 (2014), 036, 15 pages
A Note on Gluing Dirac Type Operators
on a Mirror Quantum Two-Sphere
Slawomir KLIMEK † and Matt MCBRIDE ‡
† Department of Mathematical Sciences, Indiana University-Purdue University Indianapolis,
402 N. Blackford St., Indianapolis, IN 46202, USA
E-mail: sklimek@math.iupui.edu
‡ Department of Mathematics, University of Oklahoma, 601 Elm St., Norman, OK 73019, USA
E-mail: mmcbride@math.ou.edu
Received September 30, 2013, in final form March 25, 2014; Published online March 29, 2014
http://dx.doi.org/10.3842/SIGMA.2014.036
Abstract. The goal of this paper is to introduce a class of operators, which we call quantum
Dirac type operators on a noncommutative sphere, by a gluing construction from copies of
noncommutative disks, subject to an appropriate local boundary condition. We show that
the resulting operators have compact resolvents, and so they are elliptic operators.
Key words: Dirac type operator; quantum space, C∗-algebra
2010 Mathematics Subject Classification: 46L99; 47B99; 81R60
1 Introduction
This paper is a continuation of our study of operators that we are constructing using noncom-
mutative geometry concepts, and which we named quantum Dirac type operators on noncom-
mutative spaces, see [3, 10, 11, 12, 13]. The goal of our previous papers was to provide simple
examples of such operators on noncommutative compact manifolds with boundary and then
study non-local, Atiyah–Patodi–Singer type boundary conditions and the corresponding index
problem. In particular we concentrated our efforts on showing that those operators are invertible
or invertible modulo compact operators, and that the inverses/parametrices are also compact
operators, which are the essential properties of ellipticity.
In this paper we define an interesting class of operators which we call as quantum Dirac type
operators on a noncommutative sphere of [7], by gluing two operators from noncommutative
disks, subject to an appropriate boundary condition. In doing so we are using the concepts
and techniques from our previous papers, even though in this note we employ a local boundary
condition. Other constructions of particular Dirac operators on (Podleś) quantum spheres are
described in [2, 4, 5, 6], using different techniques. It has to be stressed that in this paper we do
not single out any one Dolbeault or Dirac operator, but rather we provide a functional analytic
framework for studying a large class of such operators, an even more general class than we
introduced in [13]. We use the terminology “quantum Dirac type operators” because of many
similarities between them and classical Dirac type operators, including their structure in the
Fourier transform, as seen by comparing the formulas (2.3) and (3.3). It is a very interesting
problem to cast those operators in the spectral triples framework and we plan to address it in
future publications. The main goal, as in the past papers, is to show that our quantum Dirac
type operators have compact resolvents. We try to be quite self-contained in the presentation
so that this paper can be read without referring too much to our previous work.
In Section 2 we describe how to glue together two d-bar operators on the unit disk to make
the d-bar operator on the two-sphere in classical differential geometry. The purpose here is
mailto:sklimek@math.iupui.edu
mailto:mmcbride@math.ou.edu
http://dx.doi.org/10.3842/SIGMA.2014.036
2 S. Klimek and M. McBride
to understand the boundary condition which needs to be incorporated in the gluing. This is
followed by the computation of the kernel of the d-bar operator and the construction of its
parametrix. The calculations then lead to an alternative way of proving the standard results on
Fredholm theory and the resolvent of the d-bar operator on the sphere.
In Section 3 we define the quantum two-sphere of [7], the quantum disk of [8], our class of
quantum Dirac type operators, and the Hilbert spaces playing the role of the L2 spaces of the
previous section. Just like in the commutative case, the boundary condition will come from
how the quantum two-sphere is formed. The calculations of the kernel and the parametrix for
the quantum Dirac type operator parallel the classical case and are quite similar to those of
our previous papers. The proof of the main result on compactness of the parametrices follows.
Finally at the end of the section a specific example of the so-called q-weights is given that is
motivated by [10].
2 Gluing disks classically
In this section we describe the classical differential geometry needed to create the two-sphere and
the d-bar operator on it via gluing. We consider the standard, round two-sphere or, equivalently,
the Riemann extended plane equipped with the Fubini–Study metric, i.e. the constant curvature
metric, see [1] for details. Let
D = {z ∈ C : |z| ≤ 1}
be the disk with radius one. Identify the outside of D with D via
z 7→ 1
z
,
and so the Riemann sphere becomes the union of two unit disks modulo the gluing on the
boundary.
The Hilbert space of functions on the sphere is then identified with the direct sum
H1 = L2(D, dµ)⊕ L2(D, dµ),
where the measure dµ is given by the formula
dµ =
1
(1 + |z|2)2
d2z
with d2z being the Lebesgue measure. The Hilbert space of (0, 1) differential forms is
H2 = L2
(
D, d2z
)
⊕ L2
(
D, d2z
)
.
The geometric d-bar operator takes functions into (0, 1) differential forms: f → (∂f)dz̄ =
(∂f/∂z)dz̄. The d-bar operator D : H1 → H2 on the sphere is defined naturally to be
D(u, v) = (∂u, ∂v) (2.1)
for (u, v) ∈ H1. The domain of D is given by
dom(D) =
{
(u, v) ∈ H1 : u, v ∈ H1(D) and u(z) = v(1/z) when |z| = 1
}
, (2.2)
where H1(D) is the first Sobolev space on the disk.
A Note on Gluing Dirac Type Operators on a Mirror Quantum Two-Sphere 3
The key tool that we use to analyze D is the following Fourier decomposition for u ∈
L2(D, d2z)
u(z) =
∑
n≥0
u+n (r)einθ +
∑
n≥1
u−n (r)e−inθ,
and the same for v ∈ L2(D, dµ). This leads to natural decompositions of the Hilbert spaces
L2
(
D, d2z
) ∼= ⊕
n≥0
L2 ([0, 1], rdr)⊕
⊕
n≥1
L2 ([0, 1], rdr)
and
L2(D, dµ(z)) ∼=
⊕
n≥0
L2 ([0, 1], dµ(r))⊕
⊕
n≥1
L2 ([0, 1], dµ(r)) ,
where
dµ(r) :=
r
(1 + r2)2
dr.
The point is that the Fourier decomposition naturally carries forward to the quantum disk
case with little modifications. It will be useful to write the d-bar operator in the Fourier
components
∂u(z) =
1
2
∑
n≥0
((
u+n
)′
(r)− n
r
u+n (r)
)
ei(n+1)θ +
1
2
∑
n≥1
((
u−n
)′
(r) +
n
r
u−n (r)
)
e−i(n−1)θ. (2.3)
Using the Fourier decompositions of both u and v, the boundary condition u(z) = v(1/z) can
be rewritten as
u+n (1) = v−n (1) and u−n (1) = v+n (1), n ≥ 1, and u+0 (1) = v+0 (1). (2.4)
The above descriptions of the d-bar operator on the sphere and the Fourier expansion allow
us to prove the well known functional analytic properties of D.
Theorem 2.1. Let D be the Dirac operator defined by (2.1) on domain (2.2), then D is
a right invertible operator and it is left invertible modulo compacts. Moreover D has a com-
pact parametrix.
We begin by computing the kernel of D.
Proposition 2.2. The kernel of the d-bar operator D defined by (2.1) on domain (2.2) is one-
dimensional and consists of constant functions.
Proof. We need to solve the following uncoupled system
∂u = 0, ∂v = 0
subject to the boundary condition u(z) = v(1/z) when |z| = 1. Using the Fourier decomposition
in the first equation leads to
(u+n )′(r)− n
r
u+n (r) = 0 and (u−n )′(r) +
n
r
u−n (r) = 0.
The solution to the first equation is u+n (r) = c+n r
n for constants c+n . Moreover u−n (r) = c−n r
−n
for constants c−n . Similarly we have v+n (r) = d+n r
n and v−n (r) = d−n r
−n for constants d+n and d−n .
Since the solutions are required to be in H1(D) we must have c−n = d−n = 0 for all n ≥ 1. Now
we apply the boundary condition (2.4). For n ≥ 1 we obtain c+n = d−n and d+n = c−n which means
that c+n = 0 and d+n = 0. If n = 0 we get c+0 = d+0 is an arbitrary constant. This tells us that
KerD is one-dimensional and the proof is complete. �
4 S. Klimek and M. McBride
The next problem is to find an operator Q so that DQ(u, v) = (u, v). To write a formula for
the right inverse Q we need the following three integral operators. Let χ(t) = 1 for t ≤ 1 and
zero otherwise. Define
T
(n)
1 f(r) =
∫ 1
0
2rnρn−1f(ρ)ρdρ, T
(n)
2 f(r) = −
∫ 1
0
2χ
(
r
ρ
)
rn
ρn+1
f(ρ)ρdρ,
T
(n)
3 f(r) =
∫ 1
0
2χ
(ρ
r
) ρn−1
rn
f(ρ)ρdρ (2.5)
and T
(n)
1 , T
(n)
2 , and T
(n)
3 : L2([0, 1], rdr)→ L2([0, 1], dµ(r)).
Proposition 2.3. Let the operator D be given by (2.1) on domain (2.2). The operator D has
a right inverse Q, i.e. DQ(p, q) = (p, q) where Q(p, q) = (u, v) and
u = T
(0)
2 p+1 (r) +
∑
n≥1
(
T
(n)
2 p+n+1(r) + T
(n)
1 q−n−1(r)
)
einθ +
∑
n≥1
T
(n)
3 p−n−1(r)e
−inθ,
v = T
(0)
2 q+1 (r) +
∑
n≥1
(
T
(n)
2 q+n+1(r) + T
(n)
1 p−n−1(r)
)
einθ +
∑
n≥1
T
(n)
3 q−n−1(r)e
−inθ.
Moreover we have QD = I − C, where C(u, v) =
(
u+0 (1), u+0 (1)
)
.
Proof. In order to construct a parametrix of D, we need to solve the equation D(u, v) = (p, q)
with (u, v) subject to the boundary condition. This is the same as solving the uncoupled system
∂u = p, ∂v = q
subject to the condition (2.4). To solve ∂u = p we use the Fourier decomposition and the
equation reduces to
1
2
(
(u+n )′(r)− n
r
u+n (r)
)
= p+n+1(r) and
1
2
(
(u−n )′(r) +
n
r
u−n (r)
)
= p−n−1(r).
We obtain the following solutions
u+n (r) = c+n r
n − 2
∫ 1
r
(
r
ρ
)n
p+n+1(ρ)dρ and u−n (r) = 2
∫ r
0
(ρ
r
)n
p−n−1(ρ)dρ
for some constants c+n to be computed from the boundary condition. Here the limits of inte-
gration are determined so that the functions are non singular at r = 0 to be in the domain
of ∂. The solution for v+n (r) and v−n (r) are given by identical formulas with the corresponding
constants denoted by d+n . If n ≥ 1 then applying u+n (1) = v−n (1) we get
c+n = u+n (1) = v−n (1) = 2
∫ 1
0
ρnq−n−1(ρ)dρ
and similarly applying v+n (1) = u−n (1) we obtain
d+n = v+n (1) = u−n (1) = 2
∫ 1
0
ρnp−n−1(ρ)dρ.
Finally c+0 = d+0 . Comparing the above formulas with the integrals (2.5), we get
u+n (r) = T
(n)
1 q−n−1(r) + T
(n)
2 p+n+1(r), u−n (r) = T
(n)
3 p−n−1(r),
v+n (r) = T
(n)
1 p−n−1(r) + T
(n)
2 q+n+1(r), v−n (r) = T
(n)
3 q−n−1(r).
It remains to check that QD(u, v) = (I − C)(u, v), which is a direct but somewhat lengthly
computation and will be omitted. This finishes the proof. �
A Note on Gluing Dirac Type Operators on a Mirror Quantum Two-Sphere 5
An immediate corollary is that since the operator C defined in the above proposition is
a finite rank operator, D is left invertible modulo compacts.
The next goal is to show that Q is a compact operator. We first investigate the compactness
and norms of the integral operators T
(n)
1 , T
(n)
2 , and T
(n)
3 . This is summarized in the following
proposition.
Proposition 2.4. The three integral operators given by formula (2.5) are Hilbert–Schmidt ope-
rators. Moreover the Hilbert–Schmidt norms of these integral operators go to zero as n→∞.
Proof. We first assume n ≥ 1 for all of the operators and then consider the operator T
(0)
2
separately. For all the following estimates we use the fact that (1 + r2)−2 ≤ 1 for 0 ≤ r ≤ 1.
The Hilbert–Schmidt norm of T
(n)
1 is bounded by
‖T (n)
1 ‖
2
HS ≤ 4
∫ 1
0
∫ 1
0
r2n+1ρ2n−1dρdr =
1
n(n+ 1)
.
Similarly
‖T (n)
2 ‖
2
HS ≤ 4
∫ 1
0
∫ 1
r
(
r
ρ
)2n+1
dρdr = 4
(
1
4n
− 1
4n(n+ 1)
)
≤ 1
n
,
and
‖T (n)
3 ‖
2
HS ≤ 4
∫ 1
0
∫ r
0
ρ2n−1
r2n−1
dρdr =
1
n
.
Finally we have
∥∥T (0)
2
∥∥2
HS
≤ 4
∫ 1
0
∫ 1
r
r
ρ
dρdr = 4 lim
t→0+
∫ t
1
r ln rdr <∞
since r ln r is continuous at r = 0. It is clear that the Hilbert–Schmidt norms are all finite and
go to zero as n→∞. This completes the proof. �
Proof of Theorem 2.1. Proposition 2.3 shows that Q is the right inverse to D. Moreover
the comment directly after the proof of Proposition 2.3 shows that D is left invertible modulo
compact operators. We use the decomposition of Q in Proposition 2.3 and the fact that the
Hilbert–Schmidt norms of T
(n)
1 , T
(n)
2 , and T
(n)
3 go to zero as n → ∞ in Proposition 2.4, to
conclude that Q must be a compact operator as the norm limit of compact operators, its partial
sums. This is because the operators T
(n)
1 , T
(n)
2 , and T
(n)
3 act in mutually orthogonal subspaces
of the Hilbert space and so Q is an infinite direct sum of compact operators whose norms are
going to zero. This ends the proof. �
3 Quantum sphere
In noncommutative geometry the quantum spaces are defined to be specific C∗-algebras, which
play the role of the spaces of continuous functions. With this in mind we define the quantum
disk.
Definition 3.1. Let {ek} be the canonical basis for `2(N) and let U be the unilateral shift, i.e.
Uek = ek+1. Let C∗(U) be the C∗-algebra generated by U . The C∗-algebra C∗(U) is called the
quantum unit disk [8].
6 S. Klimek and M. McBride
Below we will need some results about C∗(U) which we will review now. There is a short
exact sequence
0 −→ K −→ C∗(U)
σ−→ C(S1) −→ 0,
where K is the algebra of compact operators in `2(N), and it is the commutator ideal of C∗(U).
The commutative quotient C∗(U)/K is isomorphic with C(S1) where the quotient map σ, also
called the restriction to the boundary map is given by σ(U) = eiθ ∈ C(S1) and σ(U∗) = e−iθ.
The related C∗-algebra morphism σ : C∗(U) 7→ C(S1), defined by the formulas σ(U) = e−iθ
and σ(U∗) = eiθ, is the composition of σ with the z → 1/z automorphism on the unit circle.
Let Kek = kek be the label operator. The operators (K,U) are called noncommutative polar
coordinates.
For a numerical function f(k), the diagonal operator f(K) in `2(N) belongs to C∗(U) iff
lim
k→∞
f(k) exists, and in fact we have
σ(f(K)) = lim
k→∞
f(k)I ∈ C(S1).
The same formula is true for σ.
For x ∈ C∗(U) define the following formal series of operators
xseries =
∑
n≥0
Unx+n (K) +
∑
n≥1
x−n (K)(U∗)n,
where x+n (k) = 〈ek, (U∗)nxek〉 and x−n (k) = 〈ek, xUnek〉.
Similarly to the usual theory of Fourier series, xseries determines x even though in general
the series is not norm convergent. Other types of convergence results can be obtained along the
lines of the usual Fourier analysis.
We have
σ(x) =
∑
n≥0
einθx+n (∞) +
∑
n≥1
x−n (∞)e−inθ (3.1)
and
σ(y) =
∑
n≥0
e−inθy+n (∞) +
∑
n≥1
y−n (∞)einθ, (3.2)
where f(∞) := lim
k→∞
f(k), provided the limit exists.
We now define the Hilbert space that plays the role of the space of square integrable functions
or the square integrable (0, 1) differential forms on the disk. While for the differential geometric
reason it is important to have appropriate measures in the L2 spaces of functions and one-
forms, those details are not important for the kind of functional analytic properties of the d-bar
operator investigated in this paper, as long as the measures are equivalent. For simplicity in the
quantum case we consider a general class of Dirac like operators acting in a fixed Hilbert space.
In the definition of that space we use Fourier series, and the finite volume property of the
disk is achieved using weights. Let {a(n)(k)} be a sequence of positive numbers such that the
sum s(n) :=
∞∑
k=0
1
a(n)(k)
exists and such that s(n) goes to zero as n → ∞. The first condition
reflects in a sense the finite volume of the disk. Notice that the weights studied in [10] decrease
monotonically in n for each k. In this paper we allow more general weights than in that paper.
For a formal power series
f =
∑
n≥0
Unf+n (K) +
∑
n≥1
f−n (K)(U∗)n
A Note on Gluing Dirac Type Operators on a Mirror Quantum Two-Sphere 7
we define the following weighted `2 norm
‖f‖2 =
∑
n≥0
∞∑
k=0
1
a(n)(k)
|f+n (k)|2 +
∑
n≥1
∞∑
k=0
1
a(n)(k)
|f−n (k)|2.
Let H1 be the Hilbert space whose elements are power series f such that ‖f‖ is finite. We have
the following density proposition.
Proposition 3.2. If x ∈ C∗(U), then xseries converges in H1. Moreover the map C∗(U) 3 x→
xseries ∈ H1 is continuous, one-to-one, and the image is dense in H1.
Proof. For x ∈ C∗(U) we need to estimate the norm of xseries. Notice first that if x is a finite
sum
x =
N∑
n=0
Unx+n (K) +
N∑
n=1
x−n (K)(U∗)n
then xseries = x. Since such x’s are dense in C∗(U), it suffices to estimate the norm of the finite
sums. For brevity, we assume that x has only the U∗ terms since the same argument will work
for the U terms as well. We have
‖xseries‖2H1
=
∞∑
k=0
N∑
n=1
1
a(n)(k)
|x−n (k)|2 = tr
(
N∑
n=1
1
a(n)(K)
|gn(K)|2
)
= tr
(
N∑
n=1
1
a(n)(K)
gn(K) (U∗)n Ungn(K)
)
= tr
(
N∑
n=1
1
a(n)(K)
gn(K) (U∗)n
N∑
l=1
U lgl(K)
)
= tr
((
N∑
n=1
1
a(n)(K)
gn(K)(U∗)n
)
x∗
)
≤
∣∣∣∣∣
∣∣∣∣∣
N∑
n=1
1
a(n)(K)
gn(K)(U∗)n
∣∣∣∣∣
∣∣∣∣∣
1
||x||,
where ||x||21 = tr(x∗x) is the trace class norm. Next we estimate∥∥∥∥∥
N∑
n=1
1
a(n)(K)
gn(K) (U∗)n
∥∥∥∥∥
2
1
= tr
(
N∑
n=1
1
a(n)(K)
gn(K) (U∗)n
N∑
l=1
U lgl(K)
1
a(n)(K)
)
= tr
(
N∑
n=1
1
a(n)(K)
gn(K) (U∗)n Ungn(K)
1
a(n)(K)
)
=
∞∑
k=0
N∑
n=1
1
a(n)(k)a(n)(k)
|gn(k)|2
≤ ‖xseries‖2H1
sup
n,k
(
1
a(n)(k)
)
.
Using the the summability conditions on the weights we obtain
sup
n,k
(
1
a(n)(k)
)
≤ sup
n
( ∞∑
k=0
1
a(n)(k)
)
= sup
n
(s(n)) ≤ const,
and hence we get ‖xseries‖2H1
≤ const‖xseries‖H1‖x‖. This shows the continuity of the map
C∗(U) 3 x→ xseries ∈ H1, and consequently xseries converges in H1 for every x ∈ C∗(U).
Next we show that the map C∗(U) 3 x → xseries ∈ H1 is one-to-one. Let x and y belong
to C∗(U) and suppose that xseries = yseries. This means that x+n (k) = y+n (k) for all n ≥ 0
8 S. Klimek and M. McBride
and all k, and x−n (k) = y−n (k) for all n ≥ 1 and all k. From x+n (k) = 〈ek, (U∗)nxek〉 and
x−n (k) = 〈ek, xUnek〉 it follows that all matrix coefficients of x and y are the same so we must
have x = y. Thus the map C∗(U) 3 x→ xseries ∈ H1 is one-to-one.
To prove density we define the following indicator function
δl(k) =
{
1, l = k,
0, l 6= k.
It is clear that Unδl(K) and δl(K)(U∗)n are in C∗(U), and moreover they form an orthogo-
nal basis for H1. Finally finite linear combinations of Unδl(K) and δl(K)(U∗)n form a dense
subspace of C∗(U), hence making it a dense subspace of H1. Thus the proof is complete. �
We can now start the analysis of the Dirac type operators on the quantum sphere.
Definition 3.3. The mirror quantum sphere of [7], denoted C(S2
m), is defined to be the following
C∗-algebra
C(S2
m) ∼= {(x, y) ∈ C∗(U)× C∗(U) : σ(x) = σ(y)} .
The Hilbert space of the square integrable functions on the mirror quantum sphere is defined
to be H := H1 ⊕H1.
To describe the class of the Dirac type operators that we will be working with we need the
following Jacobi type operators
A
(n)
f(k) = b(n+1)(k)
(
f(k)− c(n)+ (k)f(k + 1)
)
: `2
a(n)(N)→ `2
a(n+1)(N),
A(n)f(k) = b(n)(k)
(
f(k)− c(n)− (k − 1)f(k − 1)
)
: `2
a(n+1)(N)→ `2
a(n)(N),
where dom(A
(n)
) = {f ∈ `2
a(n)(N) : ‖A(n)
f‖a(n+1) < ∞}, dom(A(n)) = {f ∈ `2
a(n+1)(N) :
‖A(n)f‖a(n) < ∞}, and where `2
a(n)(N) is the space of sequences, {f(k)}, k ∈ N, such that∑
k
1
a(n)(k)
|f(k)|2 is finite. (In the above formula we assume f(−1) = 0.)
Notice that the above operators are in fact general one-step difference operators. For our
analytical purposes we will require the following conditions on the coefficients (the first condition
is a repeat from above):
• {a(n)(k)} is a sequence of positive numbers such that the sum s(n) :=
∞∑
k=0
1
a(n)(k)
exists
and s(n) goes to zero as n→∞.
• {b(n)(k)} is a sequence of positive numbers such that the sum t(n) :=
∞∑
k=0
a(n)(k)
b(n)(k)2
exists
and t(n) is bounded in n.
• {c(n)± (k)} are sequences of real numbers such that for all M , N , and n there is a positive
number κ independent of M , N , and n such that κ ≤
N∏
k=M
c
(n)
± (k) ≤ 1
κ . Moreover we
require that the product
∞∏
k=0
1
c
(n)
± (k)
exists for each n.
An example of such sequences coming from the theory of quantum groups is described at the
end of the paper.
A Note on Gluing Dirac Type Operators on a Mirror Quantum Two-Sphere 9
Definition 3.4. Dirac type operators on the quantum unit disk are the operators δ defined for
formal power series f by
δf = −
∑
n≥0
Un+1A
(n)
f+n (k) +
∑
n≥1
A(n−1)f−n (k)(U∗)n−1, (3.3)
with the Jacobi type operators A(n), A
(n)
defined above.
The structure of δ is chosen to mimic the d-bar operator from the previous section, and it
matches the structure of the Dirac type operators of our previous papers [3, 10, 11, 12, 13].
Fixing the sequences {a(n)(k)}, {b(n)(k)}, and {c(n)± (k)} we can now define the Dirac type op-
erator D on the mirror quantum sphere, the operator that we will be studying for the remainder
of this paper. We set
D(f, g) = (δf, δg), (3.4)
where dom(D) = {(f, g) ∈ H : ‖δf‖ <∞, ‖δg‖ <∞, σ(f) = σ(g)}.
We remark here that with some extra effort the operator above can be interpreted as acting
between different quantum line bundles, by checking the coefficients of the Fourier decomposi-
tion of f and δf . The description of the quantum line bundles on the quantum 2-sphere that
uses gluing from two disks can be found in [14], which can presumably be adapted to the mirror
quantum sphere studied here.
The following proposition is a direct consequence of formulas (3.1) and (3.2).
Proposition 3.5. The boundary condition σ(f) = σ(g) is equivalent to the following: if n ≥ 1
then f+n (∞) = g−n (∞), f−n (∞) = g+n (∞) and f+0 (∞) = g+0 (∞).
From this proposition we can rewrite the domain for D in the following way
dom(D) = {(f, g) ∈ H : ‖δf‖ <∞, ‖δg‖ <∞, such that (1) and (2) hold},
where
(1) If n ≥ 1 then f+n (∞) = g−n (∞) and f−n (∞) = g+n (∞),
(2) f+0 (∞) = g+0 (∞).
We can now state the main result of this paper.
Theorem 3.6. The Dirac type operator D, defined in equation (3.4), subject to the boundary
conditions given in Proposition 3.5, is a right invertible operator and is left invertible modulo
compact operators. Moreover D has a compact parametrix.
The theorem is proved in a sequence of steps. First we study the kernel of the Jacobi opera-
tors A(n) and A
(n)
that appear in the definition of D.
Lemma 3.7. The kernel of A(n) is trivial and for some constant α we have
KerA
(n)
=
{(
k−1∏
i=0
1
c
(n)
+ (i)
)
α
}
.
Proof. The goal is solve the following two equations A(n)f(k) = 0 and A
(n)
f(k) = 0. For the
first equation, since we take f(−1) = 0, it is easy to see inductively that f(k) = 0 for every k,
so that A(n) has no kernel.
10 S. Klimek and M. McBride
Using the formula for A
(n)
and solving recursively it is easy to see that A
(n)
f(k) = 0 produces
f(k) =
(
k−1∏
i=0
1
c
(n)
+ (i)
)
f(0).
We must show f(k) ∈ `2
a(n)(N). A simple computation yields
‖f‖2
a(n) =
∞∑
k=0
1
a(n)(k)
(
k−1∏
i=0
1
c
(n)
+ (i)
)2
|f(0)|2 ≤ s(n)|f(0)|2
κ2
<∞
because of the conditions on c
(n)
+ (k) and a(n)(k). Thus the proof is complete. �
Next we look at the nonhomogeneous equations for A(n) and A
(n)
.
Lemma 3.8. The solution to the equation A(n)f(k) = g(k) is
f(k) =
k∑
i=0
1
b(n)(i)
k−1∏
j=i
c
(n)
− (j)
g(i).
Moreover the solutions to the equation A
(n)
f(k) = −g(k) are
f(k) =
(
k−1∏
i=0
1
c
(n)
+ (i)
)
α−
∞∑
i=k
1
b(n+1)(i)
i−1∏
j=k
c
(n)
+ (j)
g(i),
where α is an arbitrary constant.
Proof. The formulas follow from simple calculations, see the proofs in Proposition 4.2 and
Proposition 4.4 in [10]. We need to show that f(k) in each case belongs to its respective `2
space. For the solution f(k) of A(n)f(k) = g(k) we estimate pointwise
|f(k)| ≤ 1
κ
k∑
i=0
√
a(n)(i)
b(n)(i)
g(i)√
a(n)(i)
by using the condition on c
(n)
− (k). The Cauchy–Schwarz inequality implies
|f(k)|2 ≤ 1
κ2
k∑
i=0
a(n)(i)
b(n)(i)2
k∑
i=0
|g(i)|2
a(n)(i)
≤ t(n)
κ2
‖g‖2
a(n+1) ,
and thus we have
‖f‖2
a(n) =
∞∑
k=0
1
a(n)(k)
|f(k)|2 ≤ t(n)
κ2
‖g‖2
a(n+1)
∞∑
k=0
1
a(n)(k)
=
t(n)s(n)
κ2
‖g‖2
a(n+1) <∞
by the assumptions on s(n) and t(n). A similar argument shows that the f(k) solving A
(n)
f(k) =
−g(k) is in `2
a(n+1)(N). Thus this completes the proof. �
The above lemma allows us to study the limits at infinity.
Lemma 3.9. Let f ∈ dom(A(n)), then f(∞) := lim
k→∞
f(k) exists. Moreover if f ∈ dom(A
(n)
),
then f(∞) := lim
k→∞
f(k) exists.
A Note on Gluing Dirac Type Operators on a Mirror Quantum Two-Sphere 11
Proof. If f ∈ dom(A(n)), then by using Lemma 3.8 we have
f(∞) =
∞∑
i=0
1
b(n)(i)
∞∏
j=i
c
(n)
− (j)
A(n)f(i).
Similarly, if f ∈ dom
(
A
(n))
then f(∞) =
( ∞∏
i=0
1/c
(n)
+ (i)
)
α. Because of the conditions on b(n)(k)
and c
(n)
− (k) it is clear that the infinite products and sums in the above formulas are convergent.
Thus the proof is complete. �
It will be advantageous to write the solution to A
(n)
f(k) = −g(k) in a slightly different way.
This will not only make the estimates simpler, it will also mimic the continuous case theory very
closely. Using the above we get
f(k) =
( ∞∏
i=k
c
(n)
+ (i)
)
f(∞)−
∞∑
i=k
1
b(n+1)(i)
i−1∏
j=k
c
(n)
+ (j)
g(i).
Now we look at the kernel of D.
Proposition 3.10. The kernel of D subject to the boundary condition given in Proposition 3.5,
is one-dimensional.
Proof. We wish to study the equation D(x, y) = 0, with the Fourier decomposition for x =∑
n≥0
Unx+n (K) +
∑
n≥1
x−n (K)(U∗)n, and similarly for y. This translates to the following system of
equations
δx = 0, δy = 0
subject to the boundary condition. Using the Fourier decomposition and Lemma 3.7, it follows
that x−n (k) = 0 for all k and
x+n (k) =
( ∞∏
i=k
c
(n)
+ (i)
)
x+n (∞).
We also have identical formulas for y. Since x−n (k) = 0 and y−n (k) = 0 for all k it follows that
x−n (∞) = 0 and y−n (∞) = 0. If n ≥ 1 then from the boundary condition we get 0 = y−n (∞) =
x+n (∞) which implies that x+n (k) = 0 for all k. The same type of argument shows that y+n (k) = 0
for all k. If n = 0 we get
x+0 (k) =
( ∞∏
i=k
c
(n)
+ (i)
)
x+0 (∞) and y+0 (k) =
( ∞∏
i=k
c
(n)
+ (i)
)
y+0 (∞).
However we have x+0 (∞) = y+0 (∞) which implies in particular, that x+0 (0) = y+0 (0) is an arbitrary
constant. Therefore it follows that KerD is one-dimensional. �
The next goal is to construct the parametrix of the Dirac type operator subject to the
boundary condition. First we define three integral operators that the parametrix will decompose
into in a similar way to the continuous case. The operators are
T
(n)
1 f(k) =
( ∞∏
i=k
c
(n)
+ (i)
) ∞∑
i=0
a(n−1)(i)
b(n−1)(i)
∞∏
j=i
c
(n−1)
− (j)
f(i)
a(n−1)(i)
,
12 S. Klimek and M. McBride
T
(n)
2 f(k) = −
∞∑
i=0
a(n+1)(i)
b(n+1)(i)
χ
(
k
i
)i−1∏
j=k
c
(n)
+ (j)
f(i)
a(n+1)(i)
,
T
(n)
3 f(k) =
∞∑
i=0
a(n−1)(i)
b(n−1)(i)
χ
(
i
k
)k−1∏
j=i
c
(n−1)
− (j)
f(i)
a(n−1)(i)
, (3.5)
where T
(n)
1 , T
(n)
3 : `2
a(n−1)(N) → `2
a(n)(N) and T
(n)
2 : `2
a(n+1)(N) → `2
a(n)(N) and again χ(t) = 1 for
t ≤ 1 and zero otherwise.
We will also need the following rank one operator in H defined by C(x, y) = (x̃(K), x̃(K)),
where
x̃(k) =
(
k−1∏
i=0
1
c
(n)
+ (i)
)
x+0 (0)
with x =
∑
n≥0
Unx+n (K) +
∑
n≥1
x−n (K)(U∗)n.
Proposition 3.11. The Dirac type operator D, defined in equation (3.4), subject to the boundary
condition given in Proposition 3.5, is right invertible, with the right inverse Q given by Q(p, q) =
(x, y) where
x = T
(0)
2 p+1 (k) +
∑
n≥1
Un
(
T
(n)
2 p+n+1(k) + T
(n)
1 q−n−1(k)
)
+
∑
n≥1
T
(n)
3 p−n−1(k)(U∗)n,
y = T
(0)
2 q+1 (k) +
∑
n≥1
Un
(
T
(n)
2 q+n+1(k) + T
(n)
1 p−n−1(k)
)
+
∑
n≥1
T
(n)
3 q−n−1(k)(U∗)n.
Moreover we have QD = I − C.
Proof. The goal is to solve the equation (δx, δy) = (p, q). Using the Fourier decompositions, the
equation δx = p will produce two type of equations: A
(n)
x+n (k) = −p+n+1(k) and A(n−1)x−n (k) =
q−n−1(k). Using Lemma 3.8, we immediately get
x+n (k) =
( ∞∏
i=k
c
(n)
+ (i)
)
x+n (∞)−
∞∑
i=k
1
b(n+1)(i)
i−1∏
j=k
c
(n)
+ (j)
p+n+1(i),
x−n (k) =
k∑
i=0
1
b(n−1)(i)
k−1∏
j=i
c
(n−1)
− (j)
p−n−1(i),
and the formulas for y+n (k) and y−n (k) are the same except we need to replace the p’s with q’s.
Next we apply the boundary condition for n ≥ 1. Using x+n (∞) = y−n (∞) we get
x+n (∞) =
∞∑
i=0
1
b(n−1)(i)
∞∏
j=i
c
(n−1)
− (j)
q−n−1(i).
Similarly applying y+n (∞) = x−n (∞) we obtain
y+n (∞) =
∞∑
i=0
1
b(n−1)(i)
∞∏
j=i
c
(n−1)
− (j)
p−n−1(i).
A Note on Gluing Dirac Type Operators on a Mirror Quantum Two-Sphere 13
Using the integral formulas (3.5), it’s not too hard to see that we get
x+n (k) = T
(n)
1 q−n−1(k) + T
(n)
2 p+n+1(k), x−n (k) = T
(n)
3 p−n−1(k),
y+n (k) = T
(n)
1 p−n−1(k) + T
(n)
2 q+n+1(k), y−n (k) = T
(n)
3 q−n−1(k).
The above formulas for the coefficients imply the formulas for x and y. From the construction
we have DQ(p, q) = (p, q). As with the continuous case, a direct substitution using the boundary
conditions shows that QD(x, y) = (I − C)(x, y). Thus the proof is complete. �
It follows immediately from the above proof that the C defined in the above proposition is
a finite rank operator making D left invertible modulo compacts. The next proposition is the
key step to show that the Q we just constructed in the quantum case is a compact operator.
Proposition 3.12. The three integral operators given in equation (3.5) are Hilbert–Schmidt
operators. Moreover the Hilbert–Schmidt norms of these operators go to zero as n→∞.
Proof. We first look at the Hilbert–Schmidt norm of T
(n)
3 . We have
‖T (n)
3 ‖
2
HS =
∞∑
k=0
∞∑
i=0
1
a(n)(k)
1
a(n−1)(i)
a(n−1)(i)
b(n−1)(i)
k−1∏
j=i
c
(n−1)
− (j)
2
χ
(
i
k
)
≤ 1
κ2
∞∑
k=0
1
a(n)(k)
∞∑
i=0
a(n−1)(i)
b(n−1)(i)2
=
s(n)t(n− 1)
κ2
.
The same arguments work for T
(n)
1 and T
(n)
2 leading to
‖T (n)
1 ‖HS ≤
√
s(n)t(n− 1)
κ2
and ‖T (n)
2 ‖HS ≤
√
s(n)t(n+ 1)
κ
.
Since s(n) goes to zero as n→∞ and t(n) is bounded in n, it follows that all the Hilbert–Schmidt
norms go to zero as n→∞. Thus this completes the proof. �
Proof of Theorem 3.6. We repeat the steps of the proof of Theorem 2.1. Namely Propo-
sition 3.11 shows that Q is the right inverse to D. Moreover the comment directly after the
proof of Proposition 3.11 shows that D is left invertible modulo compacts. From the decompos-
tion of Q in Proposition 3.11, and from the fact that the Hilbert–Schmidt norms of T
(n)
1 , T
(n)
2 ,
and T
(n)
3 go zero as n→∞ by Proposition 3.12, it follows that Q must be a compact operator
because it is an infinite direct sum of compact operators with norms going to zero. This ends
the proof. �
We end this paper with an example of a Dirac type operator on the quantum sphere obtained
by gluing the quantum d-bar operator of [10] with weights of [9]. Let 0 ≤ q < 1. Define the
q-weight w(k) by w2(k) = 1− qk+1, see [9]. It is clear that these weights monotonically increase
to 1. Next let UW be the weighted shift with weight w(k), in other words, UW ek = w(k)ek+1,
and define S(k) = w2(k) − w2(k − 1), so that S(K) = [U∗W , UW ]. It is clear that we have
∞∑
k=0
S(k) = 1. In our example the operator δ on the quantum disk is given by a commutation
with the above weighted unilateral shift: δx = S(K)−1/2[x, UW ]S(K)−1/2, see [10]. Then we
have the following decomposition of δ
δx = −
∞∑
n=0
Un+1A
(n)
x+n (k) +
∞∑
n=1
A(n−1)x−n (k)(U∗)n−1,
14 S. Klimek and M. McBride
where again x =
∞∑
n=0
Unx+n (k)+
∞∑
n=1
x−n (k)(U∗)n. The Hilbert space norm is given by the expres-
sion ‖x‖2H1
= Tr
(
S(K)1/2xS(K)1/2x∗
)
.
A direct calculation shows that c
(n)
+ (k) = w(k)/w(k + n), c
(n)
− (k) = (w(k)w(k + n))/w2(k +
n + 1), a(n)(k) = S−1/2(k)S−1/2(k + n), and b(n)(k) = a(n−1)(k)w(k + n − 1). It is clear that
s(n) is finite and s(n) =
∞∑
k=0
1
a(n)(k)
= qn/2(1− q)
∞∑
k=0
qk = qn/2. This calculation also shows that
s(n)→ 0 as n→∞.
Next we examine t(n). We have
t(n) =
∞∑
k=0
a(n)(k)
b(n)(k)2
=
(1− q)qn/2
q
∞∑
k=0
qk
1− qk+n
≤ q(n−2)/2
∞∑
k=0
qk =
q(n−2)/2
1− q
which is bounded in n.
Since w(k) is monotonically increasing, it follows that c
(n)
± (k) ≤ 1 for all k and n. We compute
∞∏
k=0
1
c
(n)
+ (k)
=
∞∏
k=0
√
1− qk+n+1
1− qk+1
=
1√
(1− q2) · · · (1− qn+1)
<∞
for all n. Similarly we get
∞∏
k=0
1
c
(n)
− (k)
=
∞∏
k=0
√
1− qk+n+2
1− qk+1
√
1− qk+n+2
1− qk+n+1
=
1√
(1− q)(1− q2) · · · (1− qn+1)
<∞
for all n. Therefore for κ = 1− q we have
κ ≤
N∏
k=M
c
(n)
± (k) ≤ 1
κ
for all M , N , and n. As a consequence all the coefficient conditions are satisfied and the
operator δ yields a well behaved Dirac type operator on the mirror quantum sphere.
Acknowledgements
The first author would like to thank W. Szymanski for interesting discussions that inspired this
paper. We would also like to acknowledge the referees for insightful comments and remarks.
References
[1] Ahlfors L.V., Complex analysis. An introduction to the theory of analytic functions of one complex variable,
3rd ed., International Series in Pure and Applied Mathematics, McGraw-Hill Book Co., New York, 1978.
[2] Brain S., Landi G., The 3D spin geometry of the quantum two-sphere, Rev. Math. Phys. 22 (2010), 963–993,
arXiv:1003.2150.
[3] Carey A.L., Klimek S., Wojciechowski K.P., A Dirac type operator on the non-commutative disk, Lett.
Math. Phys. 93 (2010), 107–125.
[4] Da̧browski L., Landi G., Paschke M., Sitarz A., The spectral geometry of the equatorial Podleś sphere,
C. R. Math. Acad. Sci. Paris 340 (2005), 819–822, math.QA/0408034.
[5] Da̧browski L., Sitarz A., Dirac operator on the standard Podleś quantum sphere, in Noncommutative Ge-
ometry and Quantum Groups (Warsaw, 2001), Banach Center Publ., Vol. 61, Polish Acad. Sci., Warsaw,
2003, 49–58, math.QA/0209048.
http://dx.doi.org/10.1142/S0129055X10004119
http://arxiv.org/abs/1003.2150
http://dx.doi.org/10.1007/s11005-010-0391-7
http://dx.doi.org/10.1007/s11005-010-0391-7
http://dx.doi.org/10.1016/j.crma.2005.04.003
http://arxiv.org/abs/math.QA/0408034
http://dx.doi.org/10.4064/bc61-0-4
http://arxiv.org/abs/math.QA/0209048
A Note on Gluing Dirac Type Operators on a Mirror Quantum Two-Sphere 15
[6] D’Andrea F., Da̧browski L., Landi G., Wagner E., Dirac operators on all Podleś quantum spheres, J. Non-
commut. Geom. 1 (2007), 213–239, math.QA/0606480.
[7] Hajac P.M., Matthes R., Szymański W., Noncommutative index theory for mirror quantum spheres,
C. R. Math. Acad. Sci. Paris 343 (2006), 731–736, math.KT/0511309.
[8] Klimek S., Lesniewski A., Quantum Riemann surfaces. I. The unit disc, Comm. Math. Phys. 146 (1992),
103–122.
[9] Klimek S., Lesniewski A., A two-parameter quantum deformation of the unit disc, J. Funct. Anal. 115
(1993), 1–23.
[10] Klimek S., McBride M., d-bar operators on quantum domains, Math. Phys. Anal. Geom. 13 (2010), 357–390,
arXiv:1001.2216.
[11] Klimek S., McBride M., A note on Dirac operators on the quantum punctured disk, SIGMA 6 (2010), 056,
12 pages, arXiv:1003.5618.
[12] Klimek S., McBride M., Global boundary conditions for a Dirac operator on the solid torus, J. Math. Phys.
52 (2011), 063518, 14 pages, arXiv:1103.4569.
[13] Klimek S., McBride M., Dirac type operators on the quantum solid torus with global boundary conditions,
in preparation.
[14] Wagner E., Fibre product approach to index pairings for the generic Hopf fibration of SUq(2), J. K-Theory
7 (2011), 1–17, arXiv:0902.3777.
http://dx.doi.org/10.4171/JNCG/5
http://dx.doi.org/10.4171/JNCG/5
http://arxiv.org/abs/math.QA/0606480
http://dx.doi.org/10.1016/j.crma.2006.09.021
http://arxiv.org/abs/math.KT/0511309
http://dx.doi.org/10.1007/BF02099210
http://dx.doi.org/10.1006/jfan.1993.1078
http://dx.doi.org/10.1007/s11040-010-9084-9
http://arxiv.org/abs/1001.2216
http://dx.doi.org/10.3842/SIGMA.2010.056
http://arxiv.org/abs/1003.5618
http://dx.doi.org/10.1063/1.3602276
http://arxiv.org/abs/1103.4569
http://dx.doi.org/10.1017/is010002019jkt110
http://arxiv.org/abs/0902.3777
1 Introduction
2 Gluing disks classically
3 Quantum sphere
References
|