A Note on Dirac Operators on the Quantum Punctured Disk
We study quantum analogs of the Dirac type operator −2z∂/∂z on the punctured disk, subject to the Atiyah-Patodi-Singer boundary conditions. We construct a parametrix of the quantum operator and show that it is bounded outside of the zero mode.
Збережено в:
| Дата: | 2010 |
|---|---|
| Автори: | , |
| Формат: | Стаття |
| Мова: | English |
| Опубліковано: |
Інститут математики НАН України
2010
|
| Назва видання: | Symmetry, Integrability and Geometry: Methods and Applications |
| Онлайн доступ: | http://dspace.nbuv.gov.ua/handle/123456789/146357 |
| Теги: |
Додати тег
Немає тегів, Будьте першим, хто поставить тег для цього запису!
|
| Назва журналу: | Digital Library of Periodicals of National Academy of Sciences of Ukraine |
| Цитувати: | A Note on Dirac Operators on the Quantum Punctured Disk / S. Klimek, M. McBride // Symmetry, Integrability and Geometry: Methods and Applications. — 2010. — Т. 6. — Бібліогр.: 9 назв. — англ. |
Репозитарії
Digital Library of Periodicals of National Academy of Sciences of Ukraine| id |
irk-123456789-146357 |
|---|---|
| record_format |
dspace |
| spelling |
irk-123456789-1463572019-02-10T01:23:19Z A Note on Dirac Operators on the Quantum Punctured Disk Klimek, S. McBride, M. We study quantum analogs of the Dirac type operator −2z∂/∂z on the punctured disk, subject to the Atiyah-Patodi-Singer boundary conditions. We construct a parametrix of the quantum operator and show that it is bounded outside of the zero mode. 2010 Article A Note on Dirac Operators on the Quantum Punctured Disk / S. Klimek, M. McBride // Symmetry, Integrability and Geometry: Methods and Applications. — 2010. — Т. 6. — Бібліогр.: 9 назв. — англ. 1815-0659 2010 Mathematics Subject Classification: 46L99; 47B25; 81R60 DOI:10.3842/SIGMA.2010.056 http://dspace.nbuv.gov.ua/handle/123456789/146357 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 |
We study quantum analogs of the Dirac type operator −2z∂/∂z on the punctured disk, subject to the Atiyah-Patodi-Singer boundary conditions. We construct a parametrix of the quantum operator and show that it is bounded outside of the zero mode. |
| format |
Article |
| author |
Klimek, S. McBride, M. |
| spellingShingle |
Klimek, S. McBride, M. A Note on Dirac Operators on the Quantum Punctured Disk Symmetry, Integrability and Geometry: Methods and Applications |
| author_facet |
Klimek, S. McBride, M. |
| author_sort |
Klimek, S. |
| title |
A Note on Dirac Operators on the Quantum Punctured Disk |
| title_short |
A Note on Dirac Operators on the Quantum Punctured Disk |
| title_full |
A Note on Dirac Operators on the Quantum Punctured Disk |
| title_fullStr |
A Note on Dirac Operators on the Quantum Punctured Disk |
| title_full_unstemmed |
A Note on Dirac Operators on the Quantum Punctured Disk |
| title_sort |
note on dirac operators on the quantum punctured disk |
| publisher |
Інститут математики НАН України |
| publishDate |
2010 |
| url |
http://dspace.nbuv.gov.ua/handle/123456789/146357 |
| citation_txt |
A Note on Dirac Operators on the Quantum Punctured Disk / S. Klimek, M. McBride // Symmetry, Integrability and Geometry: Methods and Applications. — 2010. — Т. 6. — Бібліогр.: 9 назв. — англ. |
| series |
Symmetry, Integrability and Geometry: Methods and Applications |
| work_keys_str_mv |
AT klimeks anoteondiracoperatorsonthequantumpunctureddisk AT mcbridem anoteondiracoperatorsonthequantumpunctureddisk AT klimeks noteondiracoperatorsonthequantumpunctureddisk AT mcbridem noteondiracoperatorsonthequantumpunctureddisk |
| first_indexed |
2025-07-10T23:31:34Z |
| last_indexed |
2025-07-10T23:31:34Z |
| _version_ |
1837304713831251968 |
| fulltext |
Symmetry, Integrability and Geometry: Methods and Applications SIGMA 6 (2010), 056, 12 pages
A Note on Dirac Operators
on the Quantum Punctured Disk?
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, mmcbride@math.iupui.edu
Received March 30, 2010, in final form July 07, 2010; Published online July 16, 2010
doi:10.3842/SIGMA.2010.056
Abstract. We study quantum analogs of the Dirac type operator −2z ∂
∂z on the punctured
disk, subject to the Atiyah–Patodi–Singer boundary conditions. We construct a parametrix
of the quantum operator and show that it is bounded outside of the zero mode.
Key words: operator theory; functional analysis; non-commutative geometry
2010 Mathematics Subject Classification: 46L99; 47B25; 81R60
1 Introduction
The main technical and computational part of the Atiyah, Patodi, Singer paper [1] is the initial
section containing a study of a nonlocal boundary value problem for the first order differential
operators of the form Γ( ∂
∂t + B) on the semi-infinite cylinder R+ × Y , where t ∈ R+ and B, Γ
live on the boundary Y . The novelty of the paper was the boundary condition, now called
the APS boundary condition, that involved a spectral projection of B. The authors explicitly
compute and estimate the fundamental solutions on the cylinder. This is later used to construct
a parametrix for the analogical boundary value problem on a manifold with boundary by gluing
it with a contribution from the interior, see also [2].
The present paper aims, in a special case, to reproduce such results in the noncommutative
setup of [5]. A similar but different study of an example of APS boundary conditions in the
context of noncommutative geometry is contained in [4].
This paper is a continuation of the analysis started in [3] and [9]. The goal of those articles was
to provide simple examples of Dirac type operators on noncommutative compact manifolds with
boundary and then study Atiyah–Patodi–Singer type boundary conditions and the corresponding
index problem. This was done for the noncommutative disk and the noncommutative annulus
and for two somewhat different types of operators constructed by taking commutators with
weighted shifts.
In this paper we consider such non-commutative analogs of the Dirac type operator ∂
∂t + 1
i
∂
∂ϕ
on the cylinder R+×S1, which we view as a punctured disk. Using a weighted shift, which plays
the role of the complex coordinate z on the disk, we construct quantum Dirac operators, and
analogs of the L2 Hilbert space of functions in which they act. We then consider the boundary
condition of Atiyah, Patodi, Singer. This is done in close analogy with the commutative case.
The main result of this note is that a quantum operator has an inverse which, minus the zero
mode, is bounded just like in Proposition 2.5 of [1]. In contrast with our previous papers the
analysis here is more subtle because of the noncompactness of the cylinder. In particular the
components of a parametrix are not compact operators and we use the Schur–Young inequality
?This paper is a contribution to the Special Issue “Noncommutative Spaces and Fields”. The full collection is
available at http://www.emis.de/journals/SIGMA/noncommutative.html
mailto:sklimek@math.iupui.edu
mailto:mmcbride@math.iupui.edu
http://dx.doi.org/10.3842/SIGMA.2010.056
http://www.emis.de/journals/SIGMA/noncommutative.html
2 S. Klimek and M. McBride
to estimate their norms. It is hoped that in the future such results will be needed to construct
spectral triples and a noncommutative index theory of quantum manifolds with boundary.
The paper is organized as follows. In Section 2 the classical APS result for the operator −2z ∂
∂z
on the cylinder is stated and re-proved using the Schur–Young inequality. Section 3 contains
the construction of the non-commutative punctured disk and the first type of noncommutative
analogs of the operator from the previous section. The operators here are similar to those
of [3]. Also in this section a non-commutative Fourier decomposition of the Hilbert spaces and
the operators is discussed. Section 4 contains the construction and the analysis of the Fourier
components of the parametrix and the proof of the main result. Finally in Section 5 we consider
the “balanced” versions of the quantum Dirac operators in the spirit of [9] and show how to
modify the previous arguments to estimate the parametrix.
2 Classical Dirac operator on the punctured disk
In this section we revisit the analysis of Atiyah, Patodi, Singer in the simple case of semi-infinite
cylinder R+ × S1, or equivalently a punctured disk. Using complex coordinates of the latter,
we construct a parametrix of a version of the d-bar operator and prove norm estimates on its
components by using different techniques than those in [1].
Let D∗ = {z ∈ C : 0 < |z| ≤ 1} be the punctured disk. Consider the following Dirac type
operator on D∗:
D = −2z
∂
∂z
.
In polar coordinates z = reiϕ the operator D has the following representation:
D = −r
∂
∂r
+
1
i
∂
∂ϕ
= −r
∂
∂r
+ B,
where B = 1
i
∂
∂ϕ is the boundary operator.
We wish to study D, subject to the APS boundary condition, on the Hilbert space L2(D∗, dµ)
with measure µ(z) given by the following formula:
dµ(z) =
1
2i|z|2
dz ∧ dz. (2.1)
Let us recall the APS condition. Let us define P≥0 to be the spectral projection of B
in L2(S1) onto the non-negative part of the spectrum of B. Equivalently, P≥0 is the orthogonal
projection onto span{einϕ}n≥0. Then we say that D satisfies the APS boundary condition when
its domain consists of those functions f(z) = f(r, ϕ) on D∗ which have only negative frequencies
at the boundary, see [1] and [2] for more details. More precisely:
dom(D) =
{
f ∈ L2(D∗, dµ) : Df ∈ L2(D∗, dµ), P≥0f(1, ·) = 0
}
. (2.2)
Notice that, by the change of variable, t = − ln r, the Dirac operator, D on L2(D∗, dµ), is
equivalent to the operator, ∂
∂t + 1
i
∂
∂ϕ on L2(R+ × S1), since one has:
dϕ ∧ dt =
1
2i|z|2
dz ∧ dz.
This matches the APS setup.
A Note on Dirac Operators on the Quantum Punctured Disk 3
We proceed the in the same way as in [1] by considering the spectral decomposition of the
boundary operator B, which in our case amounts to Fourier decomposition:
f(z) =
∑
n∈Z
fn(r)e−inϕ. (2.3)
This yields the following decomposition of the Hilbert space L2(D∗, dµ):
L2(D∗, dµ) =
⊕
n∈Z
(
L2
(
(0, 1],
dr
r
)
⊗
[
e−inϕ
]) ∼=⊕
n∈Z
L2
(
(0, 1],
dr
r
)
. (2.4)
Now we consider the decomposition of D and its inverse. The theorem below is a special case
of Proposition 2.5 of [1] but we supply a proof that generalizes to the noncommutative setup.
Define A
(n)
f(r) := −rf ′(r) − nf(r) on the maximal domain in L2((0, 1], dr
r ), and let A0
(n) be
the operator A
(n) but with domain {f(r) ∈ dom(A(n)) : f(1) = 0}. We have:
Theorem 1. Let D be the Dirac operator defined above on the domain (2.2). With respect to
the decomposition (2.4) one has
D ∼=
⊕
n>0
A
(n) ⊕
⊕
n≤0
A0
(n)
.
Moreover, there exists an operator Q such that DQ = I = QD, and
Q =
⊕
n∈Z
Q(n) = Q(0) + Q̃,
where Q̃ is bounded.
Proof. Staring with a function g(z) ∈ L2(D∗, dµ) we want to solve the following equation
Df(z) = g(z)
with f(z) satisfying the APS boundary condition. The Fourier decomposition (2.3) yields∑
n∈Z
(
−rf ′n(r)− nfn(r)
)
e−inϕ =
∑
n∈Z
gn(r)e−inϕ.
Therefore we must solve the differential equation −rf ′n(r)− nfn(r) = gn(r) where additionally
fn(1) = 0 for n ≤ 0. This, and the requirement that f is square integrable, assures that there is
a unique solution given by the following formula:
fn(r) = Q(n)gn(r) =
−
∫ r
0
(ρ
r
)n
gn(ρ)
dρ
ρ
, n > 0,∫ 1
r
(ρ
r
)n
gn(ρ)
dρ
ρ
, n ≤ 0.
This gives us the formula for the parametrix: Q = ⊕n∈ZQ(n). Showing QD = DQ = I is
a simple computation and is omitted.
We want to prove that Q̃ = ⊕n6=0Q
(n) is bounded. One has
‖Q̃‖ ≤ sup
n∈Z\{0}
∥∥Q(n)
∥∥.
In what follows we show that the Q(n) are uniformly bounded, in fact of order O
(
1
|n|
)
. The main
tool is the following inequality, see [7].
4 S. Klimek and M. McBride
Lemma 1 (Schur–Young inequality). Let T : L2(Y ) −→ L2(X) be an integral operator:
Tf(x) =
∫
K(x, y)f(y)dy.
Then one has
‖T‖2 ≤
(
sup
x∈X
∫
Y
|K(x, y)|dy
)(
sup
y∈Y
∫
X
|K(x, y)|dx
)
.
For negative n one can rewrite Q(n) as
Q(n)gn(r) =
∫ 1
0
K(r, ρ)gn(ρ)
dρ
ρ
with integral kernel K(r, ρ) = χ(r/ρ)(r/ρ)|n|. Here the characteristic function χ(t) = 1 for t ≤ 1
and is zero otherwise. Next we estimate:
sup
r
∫ 1
r
r|n|
ρ|n|+1
dρ = sup
r
1
|n|
(
1− r|n|
)
≤ 1
|n|
.
Similarly one has:
sup
ρ
∫ ρ
0
r|n|−1
ρ|n|
dr = sup
ρ
1
|n|
· 1
ρ|n|
· ρ|n| =
1
|n|
.
Thus one has by the Schur–Young inequality that ‖Q(n)‖ ≤ 1
|n| . A similar computation for
positive n gives ‖Q(n)‖ ≤ 1
|n| for all n 6= 0. Hence one has that Q̃ is bounded. �
Since in this note we do not attempt to go beyond the analysis on the semi-infinite cylinder,
we will simply ignore n = 0. In [1] this was not an issue as Q(0) is continuous when mapping
into an appropriate local Sobolev space.
3 Dirac operators on the quantum punctured disk
In this section we construct the non-commutative punctured disk and the quantum analog of the
Dirac operator of the previous section. In particular, a non-commutative Fourier decomposition
of that operator is discussed. Let us also mention that a version of a quantum punctured disk
was previously considered in [8].
We start with defining several auxiliary objects needed for our construction. Let `2(Z) be
the Hilbert space of square summable bilateral sequences, and let {ek}k∈Z be its canonical basis.
We need the following two operators: let U be the shift operator given by:
Uek = ek+1
and let K be the label operator defined by the following formula:
Kek = kek.
By the functional calculus, if f : Z → C, then f(K) is a diagonal operator and satisfies the
relation f(K)ek = f(k)ek.
A Note on Dirac Operators on the Quantum Punctured Disk 5
Next assume we are given a sequence {w(k)}k∈Z of real numbers with the following properties:
1) w(k) < w(k + 1);
2) lim
k→∞
w(k) =: w+ exists;
3) lim
k→−∞
w(k) = 0;
4) sup
k
w(k)
w(k − 1)
< ∞.
(3.1)
In particular we have w(k) > 0.
The function w : Z → C gives a diagonal operator w(K) as above. From this we define
the weighted shift operator Uw := Uw(K) which plays the role of a noncommutative complex
coordinate on the punctured disk.
Clearly:
Uwek = w(k)ek+1, U∗
wek = w(k − 1)ek−1.
Consider the commutator S := [U∗
w, Uw], for which one has Sek = (w2(k) − w2(k − 1))ek. If
we let S(k) := w2(k) − w2(k − 1), then we can write S = S(K). Notice that S is a trace class
operator and a simple computation gives tr(S) = w2
+.
The quantum punctured disk C∗(Uw) is defined to be the C∗-algebra generated by Uw.
General theory, see [6], gives us the following short exact sequence:
0 −→ K −→ C∗(Uw) σ−→ C
(
S1
)
−→ 0,
where K is the ideal of compact operators and σ is the noncommutative “restriction to the
boundary” map.
Let b ∈ C∗(Uw) and we consider the densely defined weight on C∗(Uw) by
τ(b) = tr
(
S
(
U∗
wUw
)−1
b
)
(compare with (2.1)). We use this weight to define the Hilbert space H on which the Dirac
operator will live. This is done by the GNS construction for the algebra C∗(Uw) with respect
to τ . In other words H is obtained as a Hilbert space completion
H = (C∗(Uw), 〈·, ·〉τ = ‖ · ‖2
w),
where ‖b‖2
w = τ(bb∗).
Now we are ready to define the operator that we wish to study, the quantum analog of the
operator of the previous section. Define D by the following formula:
Db = −S−1U∗
w [b, Uw] . (3.2)
Let, as before, P≥0 be the orthogonal L2 projection onto span{einϕ}n≥0. The APS boundary
conditions on D amount to the following choice of the domain:
dom(D) =
{
b ∈ H : ‖Db‖2
w < ∞, P≥0σ(b) = 0
}
. (3.3)
There are certain subtleties in this definition which are clarified in the statement of Proposition 2
at the end of this section.
The next proposition describes a (partial) Fourier series decomposition of the Hilbert spaceH.
Define
a(k) :=
w(k)2
S(k)
,
6 S. Klimek and M. McBride
and let
`2
a(Z) =
{
{g(k)}k∈Z : ‖g‖2
a =
∑
k∈Z
a(k)−1|g(k)|2 < ∞
}
.
Now we are ready for the Fourier decomposition of H which is just like (2.4).
Proposition 1. Let H be the Hilbert space defined above. Then the formula
b =
∑
n∈Z
gn(K) (U∗)n
defines an isomorphism of Hilbert spaces⊕
n∈Z
`2
a(Z) ∼= H. (3.4)
Proof. The proof is identical to the one in [9]. In particular we have
‖b‖2
w =
∑
n∈Z
tr
(
S(K)w−2(K)|gn(K)|2
)
. �
The main reason we consider the Fourier decomposition is that it (again partially) diagonalizes
the operator D. This is the subject of the next lemma. Before stating it we need some more
notation. Consider the ratios:
c(n)(k) :=
w(k + n)
w(k)
and notice that since {w(k)} is an increasing sequence we have:
c(n)(k) = 1 for n = 0,
c(n)(k) > 1 for n > 0,
c(n)(k) < 1 for n < 0.
The coefficients c(n) are needed to define the following operators in `2
a(Z). The first is:
A
(n)
g(k) = a(k)
(
g(k)− c(n)(k)g(k + 1)
)
with domain
dom(A) =
{
g ∈ `2
a(Z) : ‖Ag‖a < ∞
}
.
Additionally consider the operator A0
(n) which is the operator A
(n) but with domain
dom
(
A0
(n)) =
{
g ∈ dom
(
A
(n)) : g∞ := lim
k→∞
g(k) = 0
}
.
The last definition makes sense since by the analysis of [9] the limit limk→∞ g(k) exists for
g ∈ dom(A). One has the following proposition, which is a quantum analog of the first part of
Theorem 1.
A Note on Dirac Operators on the Quantum Punctured Disk 7
Proposition 2. With respect to the decomposition (3.4) one has:
D ∼=
⊕
n>0
A
(n) ⊕
⊕
n≤0
A0
(n)
.
Equivalently:
Db =
∑
n>0
A
(n)
gn(K)(U∗)n +
∑
n≤0
A0
(n)
gn(K)(U∗)n,
where
b =
∑
n∈Z
gn(K) (U∗)n
Proof. The proof is a direct calculation identical to the one in [9]. �
4 Construction of the parametrix
In this section we construct and analyze in detail the inverse (= a parametrix) Q for the opera-
tor D. The construction is fairly similar to the one done in Section 4 in [9], however the norm
estimates are quite different. Somewhat surprisingly the norm estimates below hold for any
choice of sequence of weights {w(k)} satisfying (3.1).
We start with a lemma containing estimates of sums through integrals. Recall that the
sequence {w(k)} is increasing with limits at ±∞ equal, correspondingly, to w+ and 0.
Lemma 2. If f(t) is a decreasing continuous function on (0, (w+)2) then
∑
l<k
f
(
w(k)2
)
S(k) =
∑
l<k
f
(
w(k)2
)(
w(k)2 − w(k − 1)2
)
≤
∫ w2
+
w(l)2
f(t)dt, (4.1)
∑
k≤l
f
(
w(k)2
)
S(k) ≤
∫ w(l)2
0
f(t)dt, (4.2)
∑
k∈Z
f
(
w(k − 1)2
)
S(k) ≥
∫ w2
+
0
f(t)dt. (4.3)
The proof of the statements of the lemma follows from a straightforward comparison of the
Riemann sums of the left hand side with the integrals on the right hand side.
Our presentation in this section is as follows. First we discuss the kernels of the A
(n) operators
for the three cases n = 0, n > 0, n < 0. Secondly we construct the parametrices for all three
cases. Thirdly we discuss the norm estimates of the parametrices, and finally we summarize the
analysis in the main result of this paper.
Below we show that the operator D has no kernel by analyzing the terms in the decomposition
of Proposition 2.
Proposition 3. The operators A
(n) for n ≥ 0 and A0
(n) for n < 0 have no kernel.
Proof. We start with n = 0. Here c(n)(k) = 1 and it is clear that the only solution of that
A
(0)
R(0) = 0 is, up to a constant, R(0) = 1. But one has∥∥R(0)
∥∥2
a
=
∑
k∈Z
1
a(k)
=
∑
k∈Z
S(k)
w(k)2
≥ const
∑
k∈Z
S(k)
w(k − 1)2
= ∞,
8 S. Klimek and M. McBride
where we used condition 4 of (3.1) as well as (4.3) for f(t) = 1/t. Therefore R(0)(K) 6∈ `2
a(Z)
and hence A
(0) has no kernel.
Next we discuss the kernel of A
(n) when n > 0. It is not too hard to see that any element of
the kernel has to be proportional to
R(n)(k) :=
∞∏
l=k
c(n)(l) =
(w+)n
w(k)w(k + 1) · · ·w(k + n− 1)
.
The norm calculation gives
‖R(n)‖2
a =
∑
k∈Z
1
a(k)
|R(n)(k)|2 =
∑
k∈Z
1
a(k)
∞∏
l=k
|c(n)(l)|2.
Since |c(n)(l)| > 1 and
∑
k∈Z
1
a(k) = ∞, the sum above diverges and hence R(n)(K) 6∈ `2
a(Z).
Finally we discuss the kernel of the operator A0
(n) when n < 0. Yet again the kernel is
formally one dimensional and spanned by
R(n)(k) =
∞∏
l=k
c(n)(l) =
w(k + n)w(k + n− 1) · · ·w(k − 1)
(w+)−n
.
While one can easily show that R(n) ∈ `2
a(Z), one however has lim
k→∞
R(n)(k) = 1 6= 0, so this
means R(n) 6∈ dom
(
A0
(n)). Thus the result follows. �
The second portion of the discussion is the construction of the parametrices for all three cases.
Since there are no kernels (and cokernels) involved we simply compute the inverses of opera-
tors A
(n). Thus, given g(k), one needs to solve the equation A
(n)
f(k) = g(k) where additionally
we need lim
k→∞
f(k) = 0 for n ≤ 0. This is done in a similar manner to the methods in [3, 9]. In
the case when n > 0 one arrives at the following formula:
f(k) = −
∑
l<k
R(n)(k)
R(n)(l)a(l)
g(l) = −
∑
l<k
w(l) · · ·w(l + n− 1)
w(k) · · ·w(k + n− 1)
S(l)
w(l)2
g(l).
Similarly in the case n ≤ 0 one has:
f(k) =
∑
k≤l
R(n)(k)
R(n)(l)
g(l)
a(l)
=
∑
k≤l
w(k + n) · · ·w(k − 1)
w(l + n) · · ·w(l − 1)
S(l)
w(l)2
g(l).
The right hand sides of the above equation give the parametrices Q(n) for all three cases. We
thus have the following:
Q(n)g(k) = −
∑
l<k
S(l)
w(l)2
g(l) for n = 0,
Q(n)g(k) = −
∑
l<k
w(l) · · ·w(l + n− 1)
w(k) · · ·w(k + n− 1)
S(l)
w(l)2
g(l) for n > 0, (4.4)
Q(n)g(k) =
∑
k≤l
w(k + n) · · ·w(k − 1)
w(l + n) · · ·w(l − 1)
S(l)
w(l)2
g(l) for n < 0.
We summarize the above analysis in the following proposition.
A Note on Dirac Operators on the Quantum Punctured Disk 9
Proposition 4. Let Q(n) be defined by the formulas above, then we have the following
A
(n)
Q(n) = I and Q(n)A
(n) = I for n > 0;
A0
(n)
Q(n) = I and Q(n)A0
(n) = I for n ≤ 0.
Next we discuss the boundedness for the parametrices in the cases n > 0 and n < 0. The
difficulty comes for k → ∞: while the ratios of weights are always less than 1, the series∑
k∈Z
S(k)
w(k)2
is not summable and we cannot replicate the estimates of [3] and [9]. In fact the
integral operators Q(n) are not Hilbert–Schmidt. The trick is to estimate most but not all weight
ratios by one. The remaining sums, containing potentially divergent terms, are estimated by
integrals using Lemma 2. We have the following result.
Proposition 5. The operators Q(n) defined above are bounded operators in `2
a(Z) when n 6= 0.
Proof. First consider the case that n > 0. Applying the Schur–Young inequality and the
inequalities (4.1), and (4.2) one has
∥∥Q(n)
∥∥2
a
≤ sup
k
(∑
l<k
w(l) · · ·w(l + n− 1)
w(k) · · ·w(k + n− 1)
S(l)
w(l)2
)
sup
l
(∑
l<k
w(l) · · ·w(l + n− 1)
w(k) · · ·w(k + n− 1)
S(k)
w(k)2
)
≤ sup
k
(
1
w(k)
∑
l<k
S(l)
w(l)
)
sup
l
(
w(l)
∑
l<k
S(k)
w(k)3
)
≤ sup
k
1
w(k − 1)
∑
l≤k−1
S(l)
w(l)
sup
l
(
w(l)
∫ w2
+
w(l)2
t−
3
2 dt
)
≤ sup
k
(
1
w(k − 1)
∫ w(k−1)2
0
t−
1
2 dt
)
· 2 sup
l
(
1− w(l)
w+
)
≤ 2 · 2 = 4.
Thus Q(n) is bounded for n > 0. Next consider the case n < 0. Here one has quite similar
estimates:
∥∥Q(n)
∥∥2
a
≤ sup
k
∑
k≤l
w(k + n) · · ·w(k − 1)
w(l + n) · · ·w(l − 1)
S(l)
w(l)2
sup
l
∑
k≤l
w(k + n) · · ·w(k − 1)
w(l + n) · · ·w(l − 1)
S(k)
w(k)2
≤ sup
k
w(k − 1)
∑
k≤l
S(l)
w(l)2w(l − 1)
sup
l
1
w(l − 1)
∑
k≤l
S(k)
w(k)
≤
(
sup
l
w(l)
w(l − 1)
)
sup
k
(
w(k − 1)
∑
k−1<l
S(l)
w(l)3
)
sup
l
1
w(l − 1)
∑
k≤l
S(k)
w(k)
≤
(
sup
l
w(l)
w(l − 1)
)
sup
k
(
w(k − 1)
∫ w2
+
w(k−1)2
t−
3
2 dt
)
sup
l
(
1
w(l − 1)
∫ w(l)2
0
t−
1
2 dt
)
≤ 4
(
sup
l
w(l)
w(l − 1)
)2
< ∞.
Thus Q(n) is bounded for n < 0 and this completes the proof. �
Finally we put together the previous information about the parametrix Q of the Dirac ope-
rator D defined in Section 3. We state the main result of this note.
10 S. Klimek and M. McBride
Theorem 2. Let D be the operator (3.2) with domain (3.3). Then there exists an operator Q
such that QD = DQ = I. Moreover, with respect to the decomposition (3.4) one has
Q =
⊕
n∈Z
Q(n) = Q(0) + Q̃, (4.5)
where the operators Q(n) are given by (4.4) and Q̃ is bounded.
Proof. By Proposition 2 one can decompose D as
⊕
n>0 A
(n)⊕
⊕
n≤0 A0
(n) which in turn gives
the decomposition (4.5) of Q. One has that
‖Q̃‖w = sup
n6=0
∥∥Q(n)
∥∥
a
.
Then from Proposition 5, one has the following inequalities
‖Q‖2
w ≤ 4
(
sup
l
w(l)
w(l − 1)
)2
< ∞,
where the last inequality follows from the assumptions in (3.1). To see that one has DQ=QD=I
we use the decompositions of Q and D and Proposition 4. This completes the proof. �
5 The balanced quantum Dirac operators
In this section we study a version of the constructions of the previous sections that is more like
the theory of [9]. The main objects: the Hilbert space and the Dirac operator are called balanced
since in their definitions the left multiplication is not preferred over the right multiplication.
Since the results for the balanced Dirac operators are completely analogous to the “unba-
lanced” case and the proofs require only trivial modification, we simply state the main steps of
the construction. The only significant difference between the two cases are the estimates on the
components of the parametrix.
In order to avoid unnecessary complications we simply recycle the old notation. As before
the starting point is the choice of a sequence of weights {w(k)}k∈Z satisfying (3.1). The Hilbert
space H is the space of power series:
b =
∑
n∈Z
gn(K) (U∗)n
but this time with a different, balanced norm:
‖b‖2
w = tr
(
S1/2w(K)−1bb∗w(K)−1S1/2
)
=
∑
n∈Z
tr
(√
S(K)S(K + n)w−1(K)w−1(K + n)|gn(K)|2
)
.
The balanced Dirac operator is
Db = −S−1/2U∗ [b, Uw]w(K)S−1/2
with the domain:
dom(D) =
{
b ∈ H : ‖Db‖2
w < ∞, P≥0σcirc(b) = 0
}
.
A Note on Dirac Operators on the Quantum Punctured Disk 11
As before the Dirac operator splits into Fourier components. To describe them we need to
modify the coefficients of the previous sections. Actually, the coefficients
c(n)(k) :=
w(k + n)
w(k)
stay the same, but we need to change:
a(n)(k) :=
w(k)w(k + n)√
S(k)S(k + n)
.
Those are used for the following previously defined operators in `2
a(Z). The first operator is:
A
(n)
g(k) = a(n)(g(k)− c(n)(k)g(k + 1))
with domain
dom(A) =
{
g ∈ `2
a(Z) : ‖Ag‖`2a(Z) < ∞
}
,
and the second operator A0
(n) is the operator A
(n) but with domain
dom
(
A0
(n)) =
{
g ∈ dom
(
A
(n)) : g∞ := lim
k→∞
g(k) = 0
}
.
With that notation, the Proposition 2 remains true. In particular one has:
D ∼=
⊕
n>0
A
(n) ⊕
⊕
n≤0
A0
(n)
.
The problem of inverting the operator D is tackled as in the previous section. The components
of the inverse are given by formulas like (4.4) with the only modification coming from the diffe-
rent a(n) coefficients. We end up with the following expressions for the parametrices:
Q(n)g(k) = −
∑
l<k
√
S(l)S(l + n)
w(l)w(l + n)
g(l) for n = 0,
Q(n)g(k) = −
∑
l<k
w(l) · · ·w(l + n− 1)
w(k) · · ·w(k + n− 1)
√
S(l)S(l + n)
w(l)w(l + n)
g(l) for n > 0,
Q(n)g(k) =
∑
k≤l
w(k + n) · · ·w(k − 1)
w(l + n) · · ·w(l − 1)
√
S(l)S(l + n)
w(l)w(l + n)
g(l) for n < 0.
One can verify directly that for the operator Q =
⊕
n∈Z Q(n) we have QD = DQ = I. The
following is the main result of this section.
Proposition 6. The operators Q(n) defined above are bounded operators in `2
a(Z) when n 6= 0.
Proof. We use the Schur–Young inequality and follows the steps of the proof of the Proposi-
tion 5, with some modifications. We show the details for n < 0, the other case is completely
analogous.
There are two sums that we need to estimate. The first sum is:
Σn
1 (k) :=
∑
k≤l
w(k + n) · · ·w(k − 1)
w(l + n) · · ·w(l − 1)
√
S(l)S(l + n)
w(l)w(l + n)
.
12 S. Klimek and M. McBride
Using Cachy–Schwartz inequality we estimate:
Σn
1 (k) ≤
∑
k≤l
w(k + n) · · ·w(k − 1)
w(l + n) · · ·w(l − 1)
S(l)
w(l)2
1/2∑
k≤l
w(k + n) · · ·w(k − 1)
w(l + n) · · ·w(l − 1)
S(l + n)
w(l + n)2
1/2
≤
w(k − 1)
∑
k≤l
S(l)
w(l)2w(l − 1)
1/2w(k + n)
∑
k≤l
S(l + n)
w(l + n)3
1/2
.
Since the weights in the denominator are bigger than the corresponding weights in the numerator,
their ratios were estimated by one. The first term on the rights hand side of the above was
already estimated in the proof of Proposition 5. The second term is essentially the same as the
first:
sup
k
w(k + n)
∑
k≤l
S(l + n)
w(l + n)3
= sup
k
w(k)
∑
k≤l
S(l)
w(l)3
.
It follows that Σn
1 (k) is bounded uniformly in n.
The second sum in the Schur–Young inequality is
Σn
2 (l) :=
∑
k≤l
w(k + n) · · ·w(k − 1)
w(l + n) · · ·w(l − 1)
√
S(k)S(k + n)
w(k)w(k + n)
and we bound it in the same fashion as the first sum:
Σn
2 (l) ≤
∑
k≤l
w(k + n) · · ·w(k − 1)
w(l + n) · · ·w(l − 1)
S(k)
w(k)2
1/2∑
k≤l
w(k + n) · · ·w(k − 1)
w(l + n) · · ·w(l − 1)
S(k + n)
w(k + n)2
1/2
≤
1
w(l − 1)
∑
k≤l
S(k)
w(k)
1/2 1
w(l + n)
∑
k≤l
S(k + n)
w(k + n)
1/2
.
Again the first term above was already estimated in the proof of Proposition 5, and the second
term is essentially the same as the first. It follows that Σn
1 (k) is uniformly bounded. Repeating
the same steps for n > 0 gives the boundedness of Q for the balanced Dirac operator. �
References
[1] Atiyah M.F., Patodi V.K., Singer I.M., Spectral asymmetry and Riemannian geometry. I, Math. Proc.
Cambridge Philos. Soc. 77 (1975), 43–69.
[2] Booß-Bavnbek B., Wojciechowski K.P., Elliptic boundary problems for Dirac operators, Mathematics: Theo-
ry & Applications, Birkhäuser Boston, Inc., Boston, MA, 1993.
[3] Carey A.L., Klimek S., Wojciechowski K.P., Dirac operators on noncommutative manifolds with boundary,
arXiv:0901.0123.
[4] Carey A.L., Phillips J., Rennie A., A noncommutative Atiyah–Patodi–Singer index theorem in KK-theory,
arXiv:0711.3028.
[5] Connes A., Non-commutative differential geometry, Academic Press, 1994.
[6] Conway J., Subnormal operators, Research Notes in Mathematics, Vol. 51, Pitman, Boston, Mass. – London,
1981.
[7] Halmos P.R., Sunder V.S., Bounded integral operators on L2 spaces, Results in Mathematics and Related
Areas, Vol. 96, Springer-Verlag, Berlin – New York, 1978.
[8] Klimek S., Lesniewski A., Quantum Riemann surfaces. III. The exceptional cases, Lett. Math. Phys. 32
(1994), 45–61.
[9] Klimek S., McBride M., D-bar operators on quantum domains, arXiv:1001.2216.
http://dx.doi.org/10.1017/S0305004100049410
http://dx.doi.org/10.1017/S0305004100049410
http://www.arxiv.org/abs/0901.0123
http://www.arxiv.org/abs/0711.3028
http://dx.doi.org/10.1007/BF00761123
http://www.arxiv.org/abs/1001.2216
1 Introduction
2 Classical Dirac operator on the punctured disk
3 Dirac operators on the quantum punctured disk
4 Construction of the parametrix
5 The balanced quantum Dirac operators
References
|