Operator representations of a q-deformed Heisenberg algebra

Konrad Schmüdgen

Citation: J. Math. Phys. 40, 4596 (1999); doi: 10.1063/1.532989

View online: http://dx.doi.org/10.1063/1.532989
View Table of Contents: http://jmp.aip.org/resource/1/JMAPAQ/v40/i9
Published by the American Institute of Physics.

Additional information on J. Math. Phys.

Journal Homepage: http://jmp.aip.org/
Journal Information: http://jmp.aip.org/about/about_the_journal
Top downloads: http://jmp.aip.org/features/most_downloaded
Information for Authors: http://jmp.aip.org/authors

Downloaded 14 Apr 2013 to 132.174.255.3. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jmp.aip.org/about/rights_and_permissions
 JOURNAL OF MATHEMATICAL PHYSICS VOLUME 40, NUMBER 9 SEPTEMBER 1999

Operator representations of a q-deformed Heisenberg

algebra
Konrad Schmüdgena)
Universität Leipzig, Fakultät für Mathematik und Informatik und NTZ,
Augustusplatz 10/11, D-04109 Leipzig, Germany
共Received 3 June 1998; accepted for publication 9 July 1998兲
A class of well-behaved *-representations of a q-deformed Heisenberg algebra
previously introduced 关Phys. Lett. B 291, 273 共1992兲; Z. Phys. C 64, 335 共1994兲兴 is
studied and classified. © 1999 American Institute of Physics.
关S0022-2488共99兲02008-3兴

I. INTRODUCTION
The idea to develop a q-deformed quantum mechanics by using quantum groups has been
investigated in several papers.1–5 Such approaches are usually based on a q-deformed phase space
algebra which is derived from the noncommutative differential calculus of the q-deformed con-
figuration space.6,7 Following the standard procedure in quantum mechanics one has to represent
the q-deformed position and momentum operators by essentially self-adjoint operators acting on a
Hilbert space. More precisely, one has to find appropriate *-representations of the phase space
*-algebra by unbounded operators in a Hilbert space. In the case of general Euclidean or
Minykowski phase spaces the study and classification of these *-representations turns out to be
technically complicated because of the many relations and also because of the various difficulties
concerning unbounded operators.
The aim of this paper is to give a rigorous treatment of well-behaved operator representations
for one of the simplest example—the one-dimensional q-deformed Heisenberg algebra which was
invented in Refs. 4 and 2. Representations of this algebra have been investigated in Ref. 2. Since
this *-algebra occurs as a subalgebra of other larger *-algebras, the study of general not neces-
sarily irreducible *-representations seems to be important as well. We shall develop and analyze
an operator-theoretic model for such general representations of the q-deformed Heisenberg alge-
bra. This model might be used as a tool kit for the study of representations of larger *-algebras.
This paper is organized as follows. Section II contains the definition and some simple alge-
braic properties of the q-deformed Heisenberg algebra A(q). In Sec. III we develop a general
operator-theoretic model for certain triples of operators which will lead in Sec. VI to representa-
tions of the *-algebra A(q). In Sec. IV the irreducibility and the unitary equivalence of these
operator triples are investigated and a number of examples are treated. In Sec. V we give a
characterization of these operator triples by a number of natural conditions. In Sec. VI we define
the self-adjoint *-representations of the *-algebra A(q) obtained by means of these operator
triples.
In a forthcoming paper we shall study the spectrum of the operator X. For this analysis the
q-Fourier transform8,9 will play a crucial role.

II. THE q-HEISENBERG ALGEBRA

For a positive real number q⫽1, let A(q) denote the complex unital algebra with four
generators p,x,u,u⫺1 subject to the defining relations

up⫽qpu, ux⫽q ⫺1 xu, uu⫺1 ⫽u⫺1 u⫽1, 共1兲

a兲
Electronic mail: [email protected]

0022-2488/99/40(9)/4596/10/$15.00 4596 © 1999 American Institute of Physics

Downloaded 14 Apr 2013 to 132.174.255.3. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jmp.aip.org/about/rights_and_permissions
 J. Math. Phys., Vol. 40, No. 9, September 1999 Operator representations of a q-deformed . . . 4597

px⫺qxp⫽i共 q 3/2⫺q ⫺1/2兲 u, xp⫺qpx⫽⫺i共 q 3/2⫺q ⫺1/2兲 u⫺1 , 共2兲

where i denotes the imaginary unit. An equivalent set of relations is obtained if 共2兲 is replaced by

px⫽iq 1/2u⫺1 ⫺iq ⫺1/2u, xp⫽iq ⫺1/2u⫺1 ⫺iq 1/2u. 共2⬘兲

From 共1兲 and 共2兲⬘ it follows that the set of elements 兵pr un ,xs un ; r苸N0 , s苸N, n苸Z其 is a vector
space basis of A(q).
The algebra A(q) becomes a *-algebra with involution defined on the generators by

p⫽p* , x⫽x* , u* ⫽u⫺1 . 共3兲

Indeed, it suffices to check that the defining relations 共1兲 and 共2兲⬘ of A(q) are invariant under the
involution 共3兲 which is easily done.
From 共1兲, 共2兲⬘, and 共3兲 we conclude that there are *-isomorphisms ␳ 1 and ␳ 2 of the *-algebras
A(q) and A(q ⫺1 ) such that

␳ 1 共 x兲 ⫽p, ␳ 1 共 p兲 ⫽x, ␳ 1 共 u兲 ⫽u, and ␳ 2 共 x兲 ⫽x, ␳ 2 共 p兲 ⫽p, ␳ 2 共 u兲 ⫽⫺u* .

Because the *-algebras A(q) and A(q ⫺1 ) are isomorphic, we shall assume in what follows that
0⬍q⬍1.

III. AN OPERATOR-THEORETIC MODEL

共1兲 Let ␮ 1 be a finite positive Borel measure on the interval 关 q,1). The measure ␮ 1 extends
uniquely to a Borel measure ␮ on the half-axis R⫹ ⫽(0,⫹⬁) by setting ␮ (q n M)ªq n ␮ 1 (M) for
any Borel subset M of 关 q,1). Then ␮ has the property that ␮ (qN)⫽q ␮ (N) for an arbitrary Borel
subset N of R⫹ or equivalently that d ␮ (qt)/qt⫽d ␮ (t)/t for t苸R⫹ . We shall work with the
Hilbert spaces HªL 2 (R⫹ , ␮ ) and HªL 2 „关 q,1), ␮ 1 …. First we define three linear operators U, P,
and X on the Hilbert space H:
共i兲 (U f )(t)⫽q 1/2 f (qt) for f 苸H,
共ii兲 ( P f )(t)⫽t f (t) for f 苸D( P)ª 兵 f 苸H:t f (t)苸H其 ,
共iii兲 (X f )(t)⫽it ⫺1 „f (q ⫺1 t)⫺ f (qt)… for f 苸D(X)ª 兵 f 苸H:t ⫺1 f (t)苸H其 .

These operators will play a crucial role throughout this paper. Roughly speaking and ignoring
technical subtleties 共domains, boundary conditions, etc.兲, we shall show that for all ‘‘well-
behaved’’ *-representations of the q-deformed Heisenberg algebra A(q) the images of the gen-
erators u, p, and x act by the same formulas as the operators U, P, and X, respectively.
Obviously, P is an unbounded self-adjoint operator on H. Using the relation d ␮ (qt)/qt
⫽d ␮ (t)/t one easily verifies that U is a unitary operator and that X is a symmetric operator on H.
Let D0 be the set of functions f 苸H such that supp f 苸 关 a, b 兴 for some a⬎0 and b⬎0. 共Note that
a and b may depend on f.兲 Clearly, D0 is dense linear subspace of H which is invariant under U,
P, and X. It is straightforward to check that the operators P, X, and U applied to functions f
苸D0 satisfy the defining relations 共1兲, 共2兲, and 共3兲 of the *-algebra A(q). In turns out that the
symmetric operator X is not essentially self-adjoint. Our next aim is to characterize the domain of
the adjoint operator X * .
For f 苸H⫽L 2 „关 q,1), ␮ 1 … let f e and f o be the functions on R⫹ defined by

f e 共 q 2n t 兲 ⫽ f o 共 q 2n⫹1 t 兲 ⫽ f 共 t 兲 for n苸N0 , t苸 关 q,1兲 , and f e 共 t 兲 ⫽ f o 共 t 兲 ⫽0 otherwise. 共4兲

Clearly, f e and f o are in H⫽L 2 (R⫹ , ␮ ) and we have U( f e )⫺q 1/2 f 苸D(X) and U( f o )⫺q 1/2 f e
苸D(X). Let He and Ho denote the set of functions f e and f o , respectively, where f 苸H
⫽L 2 „关 q,1), ␮ 1 ….
Lemma 1: The domain D(X * ) is the direct sum of vector spaces D(X), He and Ho .

Downloaded 14 Apr 2013 to 132.174.255.3. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jmp.aip.org/about/rights_and_permissions
 4598 J. Math. Phys., Vol. 40, No. 9, September 1999 Konrad Schmüdgen

Proof: It is straightforward to check that D(X)⫹He ⫹Ho 債D(X * ). In order to prove the
converse, let g苸D(X * ). Then, by definition there is an h苸H such that 具 X f ,g 典 ⫽ 具 f ,h 典 for all f
苸D(X). Inserting the definition of X and using once more the fact that ␮ (qN)⫽q ␮ (N) for an
arbitrary Borel subset N of R⫹ we easily conclude that h(t)⫽it ⫺1 „g(q ⫺1 t)⫺g(qt)…. For a
function f 苸H let f n denote the function in L 2 „关 q,1), ␮ ⫹
1 … given by f n (t)⫽ f (q t). Then we get
n

⬁ ⬁
储 g n⫹1 ⫺g n⫺1 储 2 n
2
储 h 储 L 2共 R , ␮ 兲⫽
⫹
兺 储 h n 储 q ⭓ n⫽0
n⫽⫺⬁
兺
2 n
q 2n
q .

For n苸N we set ␣ n ª 储 g n⫹1 ⫺g n⫺1 储 q ⫺n/2. Since h苸L 2 (R, ␮ ), the sequence ( ␣ n ) is in l 2 . From
the inequality

储 g 2r ⫺g 2s 储 ⭐ ␣ 2r⫹1 q 共 2r⫹1 兲 /2⫹¯⫹ ␣ 2s⫹1 q 共 2s⫹1 兲 /2

we obtain

储 g 2r ⫺g 2s 储 2 ⭐ 冉兺 冊 ⬁

i⫽2s⫹1
兩 ␣ i 兩 2 q 2s⫹1 共 1⫺q 2 兲 ⫺1 , r⭓s. 共5兲

Since ( ␣ n )苸l 2 , this implies that the sequence (g 2n ) n苸N converges in the Hilbert space
L 2 „关 q,1), ␮ 1 …. Let us denote its limit by ␰. We extend ␰ to a function ␰ e on R⫹ by setting
␰ e (q 2n t)ª ␰ (t) and ␰ e (q 2n⫹1 t)ª0 for n苸N0 , t苸 关 q,1), and ␰ e (t)⫽0 for t⭓1. Replacing even
indices by odd indices, a similar reasoning yields functions ␨ 苸L 2 „关 q,1), ␮ 1 … and ␨ o on R⫹ such
that ␨ o (q 2n⫹1 t)⫽ ␨ (t) and ␨ o (q 2n t)⫽0 for n苸N, t苸 关 q,1), and ␨ o (t)⫽0 for t⭓1. By construc-
tion, ␰ e 苸He and ␨ o 苸Ho . Our proof is complete once we have shown that f ªg⫺ ␰ e ⫺ ␨ o belongs
to the domain D(X) of the operator X.
Letting r˜⬁ in 共5兲, we get
⬁
储 ␰ ⫺g 2s 储 2 ⭐q 2s⫹1 共 1⫺q 2 兲 ⫺1 兺
n⫽0
兩 ␣ 2n 兩 2 . 共6兲

From 共6兲 and the corresponding estimation of 储 ␨ ⫺g 2s⫹1 储 2 we obtain

⬁ ⬁ ⬁ ⬁
储 f n储 2 n 储 ␰ ⫺g 2r 储 2 储 ␨ ⫺g 2r⫹1 储 2
兺
n⫽0
储 t ⫺1 f n 共 t 兲储 2 q n ⭐ 兺
n⫽0 q 2n⫹2
q ⫽ 兺
r⫽0 q 2r⫹2
⫹
q 2r⫹3
⫽ 共 q⫺q 3 兲 ⫺1
n⫽0
兩 ␣ n 兩 2 ⬍⬁. 兺
Since f (t)⫽g(t) for t⭓1, this inequality implies that the functions t ⫺1 f (t) and f (t) are in
L 2 (R⫹ , ␮ ). Thus, f 苸D(X). 䊏
As shown in the preceding proof, for any function g苸D(X * ) the ‘‘even components’’ g 2n
and the ‘‘odd components’’ g 2n⫹1 both have ‘‘boundary limits’’ ␰ and ␨ in L 2 „关 q,1), ␮ 1 …. By
Lemma 1, any element f 苸D(X * ) is of the form f ⫽ f X ⫹ f e ⫹ f o with uniquely determined func-
tions f X 苸D(X), f e 苸He and f o 苸Ho . By the definition of He and Ho , there exist unique func-
tions f e , f o 苸H⫽L 2 „关 q,1), ␮ 1 … such that ( f e ) e ⫽ f e and ( f o ) o ⫽ f o , where the function ( f e ) e and
( f o ) o on R are given by 共4兲. This notation will be kept in the sequel.
Let 具•,•典 and 共•,•兲 denote the scalar products of the Hilbert spaces L 2 (R⫹ , ␮ ) and
L ( 关 q,1),t ⫺1 ␮ 1 ), respectively.
2

Lemma 2: For arbitrary functions f ,g苸D(X * ) we have

1
具 X * f ,g 典 ⫺ 具 f ,X * g 典 ⫽
兵 共 f ⫹ f ,g ⫹g o 兲 ⫺ 共 f e ⫺ f o ,g e ⫺g o 兲 其 . 共7兲⫹
2i e o e
Proof: Let h苸L 2 „( 关 q,1), ␮ 1 …. From the definitions of the operator X and of the functions h e ,h o
苸L 2 (R, ␮ ) we easily derive that (X * h e )(t)⫽⫺it ⫺1 h(qt) for t苸 关 1,q ⫺1 ), (X * h e )(t)⫽0 for t

Downloaded 14 Apr 2013 to 132.174.255.3. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jmp.aip.org/about/rights_and_permissions
 J. Math. Phys., Vol. 40, No. 9, September 1999 Operator representations of a q-deformed . . . 4599

苸R⫹ ⶿ 关 1,q ⫺1 ), (X * h o )(t)⫽⫺it ⫺1 h(t) for t苸 关 q,1), and (X * h o )(t)⫽0 for t苸R⫹ ⶿ 关 q,1). Insert-
ing these expressions and using the symmetry of the operator X we compute

具 X * f ,g 典 ⫺ 具 f ,X * g 典 ⫽ 具 X * f o ,g e 典 ⫺ 具 f e ,X * g o 典

⫽⫺i 冕 q
1
„f o 共 t 兲 g e 共 t 兲 ⫹ f e 共 t 兲 g o 共 t 兲 …t ⫺1 d ␮ 共 t 兲

1
⫽⫺i兵 共 f o ,g e 兲 ⫹ 共 f e ,g o 兲 其 ⫽ 兵 共 f ⫹ f ,g ⫹g o 兲 ⫺ 共 f e ⫺ f o ,g e ⫺g o 兲 其 .
2i e o e
䊏

Let us illustrate the preceding by the simplest example.

Example 1: Let ␮ 1 be the Delta measure ␦ a , where a is a fixed number from the interval
关 q,1). Then the measure ␮ is supported on the points aq n ,n苸Z, and we have ␮ ( 兵 aq n 其 )
⫽q n ␮ ( 兵 a 其 )⫽q n . Hence the scalar product of the Hilbert space H⫽L 2 (R⫹ , ␮ ) is given by the
Jackson integral
⫹⬁

具 f ,g 典 ⫽ 兺
n⫽⫺⬁
f 共 aq n 兲 g 共 aq n 兲 q n .

Let e n 苸H be the function e n (t)⫽q ⫺n/2␦ aq n , where ␦ st is the usual Kronecker symbol. Then the
t

vectors e n ,n苸Z, form an orthonormal basis of H and the actions of the operators U, P, and X on
these vectors are given by

i
Ue n ⫽e n⫺1 , Pe n ⫽aq n e n , Xe n ⫽ 共 q ⫺1/2e n⫹1 ⫺q 1/2e n⫺1 兲 .
aq n

These equations are in accordance with formulas 共5兲 in Ref. 2. If f is the function in
L 2 „关 q,1), ␮ 1 …⬵C with f (a)⫽1, then by definition f e (aq 2n )⫽ f o (aq 2n⫹1 )⫽1, f e (aq 2n⫹1 )
⫽ f o (aq 2n )⫽0 for n苸N0 , and f e (t)⫽ f o (t)⫽0 for t⭓1. Then we have D(X * )⫽D(X)⫹C• f e
⫹C• f o by Lemma 1 and formula (7) ⫹ reads as

具 X * 共 ␸ ⫹ ␣ 1 f e ⫹ ␤ 1 f o 兲 , ␺ ⫹ ␣ 2 f e ⫹ ␤ 2 f o 典 ⫺ 具 ␸ ⫹ ␣ 1 f e ⫹ ␤ 1 f o ,X * 共 ␺ ⫹ ␣ 2 f e ⫹ ␤ 2 f o 兲 典
⫽⫺ia ⫺1 兵 ␤ 1 ¯␣ 2 ⫹ ␣ 1 ¯␤ 2 其
1
⫽ 兵 共 ␣ 1 ⫹ ␤ 1 兲共 ␣ 2 ⫹ ␤ 2 兲 ⫺ 共 ␣ 1 ⫺ ␤ 1 兲共 ␣ 2 ⫺ ␤ 2 兲 其
2ia

for ␣ 1 , ␤ 1 , ␣ 2 , ␤ 2 苸C. 䊏
共2兲 The above considerations carry over almost verbatim to the case where the positive
half-axis R⫹ is replaced by the negative half-axis R⫺ ⫽(⫺⬁,0). Any positive finite Borel measure
␮ 1 on the interval 关 q,1) induces a positive Borel measure ␮ on R⫺ by defining ␮ (⫺q n M)
ªq n ␮ 1 (M) for a Borel subset M of 关 q,1). The operators U, P, and X on the Hilbert space
H⫺ ªL 2 (R⫺ , ␮ ) are defined by the same formulas as in the preceding subsection and Lemma 1
and its proof remain valid in this case as well. However, there is an essential difference which will
be crucial in the sequel: Since in the proof of Lemma 2 the integration is over the interval
(⫺1,⫺q 兴 , the expression on the right-hand side of (7) ⫹ must be multiplied by ⫺1. That is,
instead of (7) ⫹ we now have

1
具 X * f ,g 典 ⫺ 具 f ,X * g 典 ⫽ 兵 共 f ⫹ f ,g ⫹g o 兲 ⫺ 共 f e ⫺ f o ,g e ⫺g o 兲 其 共7兲⫺
2i e o e
for f ,g苸D(X * ).

Downloaded 14 Apr 2013 to 132.174.255.3. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jmp.aip.org/about/rights_and_permissions
 4600 J. Math. Phys., Vol. 40, No. 9, September 1999 Konrad Schmüdgen

共3兲 After the preceding preparations we are now able to develop the operator-theoretic model
for the description of *-representations of the q-Heisenberg algebra A(q). For this let us fix two
families 兵 ␮ 1j,⫹ ; j苸I ⫹ 其 and 兵 ␮ 1j,⫺ ; j苸I ⫺ 其 of finite positive Borel measures on the interval 关 q,1).
As above, we define the Hilbert spaces H j,⫾ ªL 2 (R⫾ , ␮ j,⫾ ), j苸I ⫾ , and the operators U j,⫾ ,
P j,⫾ , and X j,⫾ acting therein. We shall work with the representation Hilbert space H⫽H⫹
丣 H⫺ , where H⫹ ª 丣 j苸I H j,⫹ and H⫺ ª 丣 j苸I H j,⫺ . The elements of H are pairs f
⫹ ⫺
⫽(f⫹ ,f⫺ ), where f⫹ ⫽( f j,⫹ ; j苸I ⫹ )苸H⫹ and f⫺ ⫽( f j,⫺ ; j苸I)苸H⫺ . Let U, P, and X denote the
operators on H which are defined as the direct sums of the operators U j,⫹ ,U j,⫺ ; P j,⫹ , P j,⫺ ; and
X j,⫹ ,X j,⫺ , respectively. Clearly, U is a unitary operator and P is a self-adjoint operator on H.
The operator X is only symmetric, but not self-adjoint. Our next aim is to describe all self-adjoint
extensions X̃ of X on H which have the property that UX̃U ⫺1 ⫽qX̃.
Let V and W be two unitary linear transformations of the Hilbert space H⫺
ª 丣 j苸I ⫺ L 2 „关 q,1),t ⫺1 ␮ 1j,⫺ … on the Hilbert space H⫹ ª 丣 j苸I ⫹ L 2 „关 q,1),t ⫺1 ␮ 1j,⫹ …. We define a
linear operator X V,W as being the restriction of the adjoint operator X * to the domain

D共 X V,W 兲 ª 兵 f⫽fX ⫹fe ⫹fo 苸D共 X * 兲 :fX 苸D共 X 兲 ,

共8兲
f⫹ ⫺ ⫺ ⫺ ⫺ ⫹ ⫹ ⫺ ⫺ ⫺
e ⫽V 共 fe ⫹f0 兲 ⫹W 共 fe ⫺f0 兲 , f0 ⫽V 共 fe ⫹f0 兲 ⫺W 共 fe ⫺f0 兲 .

Proposition 1: X V,W is a self-adjoint operator on H such that X債X V,W and UX V,W U *
⫽qX V,W . In particular, we have UD(X V,W )⫽D(X V,W ). Conversely, for any self-adjoint exten-
sion X̃ of X satisfying UD(X̃)債D(X̃) there exist unitary transformations V,W of H⫹ onto H⫺
such that X̃⫽X V,W .
Proof: From (7) ⫹ and (7) ⫺ we obtain

⫺2i共 具 X * f,g典 ⫺ 具 f,X * g典 兲 ⫽ 共 f⫹ ⫹ ⫹ ⫹ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺

e ⫹fo ,ge ⫹go 兲 ⫹ 共 fe ⫺fo ,ge ⫺go 兲 ⫺ 共 fe ⫹fo ,ge ⫹go 兲

⫺ 共 f⫹ ⫹ ⫹ ⫹
e ⫺fo ,ge ⫺go 兲 共9兲

for arbitrary elements f⫽fX ⫹fe ⫹fo and g⫽gX ⫹ge ⫹go of D(X * ). Here f⫹ e denotes the sequence
( f ej,⫹ ; j苸I ⫹ )苸H⫹ with f ej,⫹ 苸L 2 „关 q,1), ␮ 1j … such that the extension ( f ej,⫹ ) e of f ej,⫹ to R⫹ by
means of formula 共4兲 is just the ( j,⫹)-component of the vector fe 苸H. A similar meaning at-
tached to the other symbols f ⫺ ⫹ ⫺ ⫹ ⫺ ⫹ ⫺
e , f o , f o ,ge ,go ,go ,go occuring in 共9兲. If f,g苸D(X V,W ), then we
have f e ⫹fo ⫽V(fe ⫹fo ), ge ⫹g⫺
⫹ ⫺ ⫺ ⫺ ⫹ ⫺ ⫺ ⫹ ⫹ ⫺ ⫺
o ⫽V(ge ⫹go ), fe ⫺fo ⫽W(fe ⫺fo ), and ge ⫺go ⫽W(ge
⫹ ⫹ ⫺
⫺
⫺go ) by 共8兲. Since X V,W 債X * , we therefore obtain that 具 X V,W f,g典 ⫺ 具 f,X V,W g典 ⫽0 by 共9兲, that is,
the operator X V,W is symmetric. Now let g苸D„(X V,W ) * …. Since X債X V,W 債(X V,W ) * 債X * , we
then have 具 X * f,g典 ⫽ 具 f,X * g典 and hence

共 f⫹ ⫹ ⫹ ⫹ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫹ ⫹ ⫹ ⫹
e ⫹fo ,ge ⫹go 兲 ⫹ 共 fe ⫺fo ,ge ⫺go 兲 ⫽ 共 fe ⫹fo ,ge ⫹go 兲 ⫹ 共 fe ⫺fo ,ge ⫺go 兲 共10兲

for all f苸D(X V,W ) by 共9兲. Inserting 共8兲 into 共10兲, we get

„f ⫺ ⫺ ⫹ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫹ ⫹
e ⫹fo ,V * 共 ge ⫹go 兲 …⫹ 共 fe ⫺fo ,ge ⫺go 兲 ⫽ 共 fe ⫹fo ,ge ⫹go 兲 ⫹„fe ⫺fo ,W * 共 ge ⫺go 兲 ….
共11兲

From the construction it is clear that for arbitrary h,k苸H⫺ there exists f苸D(X V,W ) such that f⫺e
⫹f⫺ ⫺ ⫺ ⫹ ⫹
o ⫽h and fe ⫺fo ⫽k. Therefore, it follows from 共11兲 that V * (ge ⫹go )⫽ge ⫹go and
⫹ ⫺

W * (g⫹ ⫹ ⫺ ⫺
e ⫺go )⫽ge ⫺go , which in turn implies that g苸D(X V,W ). Thus we have shown that the
operator X V,W is self-adjoint. From the relations U(fe )⫺q 1/2fo 苸D(X) and U(fo )⫺q 1/2fe 苸 ⳵ (X)
we see that UD(X V,W )⫽D(X V,W ). Since UXU * ⫽qX and hence UX * U * ⫽qX * and X V,W is the
restriction of X * to D(X V,W ), the latter yields UX V,W U * ⫽qX V,W .

Downloaded 14 Apr 2013 to 132.174.255.3. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jmp.aip.org/about/rights_and_permissions
 J. Math. Phys., Vol. 40, No. 9, September 1999 Operator representations of a q-deformed . . . 4601

Conversely, suppose that X̃ is a self-adjoint extension of X such that UD(X̃)債D(X̃). Since X̃

is symmetric, we have Eq. 共10兲 for arbitrary elements f,g苸D(X̃). By assumption, Uf苸D(X̃) for
all f苸D(X̃). Replacing f by Uf in 共10兲 we get

共 f⫹ ⫹ ⫹ ⫹ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫹ ⫹ ⫹ ⫹
e ⫹fo ,ge ⫹go 兲 ⫹ 共 fo ⫺fe ,ge ⫺go 兲 ⫽ 共 fe ⫹fo ,ge ⫹go 兲 ⫹ 共 fo ⫺fe ,ge ⫺go 兲 . 共12兲

Setting f⫽g and combining formulas 共10兲 and 共12兲 we obtain

储 f⫹ ⫹ ⫺ ⫺
e ⫹fo 储 ⫽ 储 fe ⫹fo 储 and 储 f⫹ ⫹ ⫺ ⫺
e ⫺fo 储 ⫽ 储 fe ⫺fo 储 共13兲

for all f苸D(X̃).

For f苸D(X * ) we abbreviate B ⫾ (f)⫽(f⫾ ⫾ ⫾ ⫾
e ⫹f0 ,fe ⫺fo ). The vector space B ⫾ (X̃)⫽ 兵 B ⫾ (f):f
苸D(X̃) 其 is called the ‘‘boundary space’’ of the operator X̃. We shall show that B ⫹ (X̃)⫽H⫹
丣 H⫹ and B ⫺ (X̃)⫽H⫺ 丣 H⫺ . First let us note that the spaces B ⫾ (X̃) are closed in H⫾ 丣 H⫾ .
Otherwise, let X5 denote the restriction of X * to the domain D(X5 )⫽ 兵 f苸D(X * ):B ⫾ (f)
苸B ⫾ (X̃) 其 , where the bar means the closure in the Hilbert space H⫾ 丣 H⫾ . The symmetry of an
operator Y such that X債Y 債X * is equivalent to the validity of Eq. 共10兲 for all f,g苸D(Y ). Hence
X5 is symmetric, because X̃ is so. Since a self-adjoint operator has no proper symmetric extension,
we conclude that X̃⫽X5 which means that B ⫹ (X̃) and B ⫺ (X̃) are closed. Next let us suppose that
( ␰ , ␨ )⬜B ⫹ (X̃) in H⫹ 丣 H⫹ . We then choose a vector g苸D(X̃) such that ␰ ⫽g⫹ ⫹
e ⫹go , ␨ ⫽ge
⫹
⫹ ⫺ ⫺
⫺go and ge ⫽g0 ⫽0. Then the right-hand side of 共9兲 vanishes for all f苸D(X̃), so that 具 X̃f,g典
⫽ 具 X * f,g典 ⫽ 具 f,X * g典 for all f苸D(X̃) by 共9兲. Consequently, g苸D(X̃ * ). Since X̃ is self-adjoint, g
must be in D(X̃). Because ( ␰ , ␨ )⬜B ⫹ (X̃), this implies that ␰ ⫽ ␨ ⫽0. This proves that B ⫹ (X̃)
⫽H⫹ 丣 H⫹ . Similarly B ⫺ (X̃)⫽H⫺ 丣 H⫺ .
Since B ⫾ (X̃)⫽H⫾ 丣 H⫾ as just shown, it follows from 共13兲 that there are unitary operators V
and W of H⫺ onto H⫹ such that f⫹ ⫹ ⫺ ⫺ ⫹ ⫹ ⫺ ⫺
e ⫹fo ⫽V(fe ⫹fo ) and fe ⫺fo ⫽W(fe ⫺fo ) for all f苸D(X̃).
That is, D(X̃)債D(X V,W ). Since X̃ and X V,W are self-adjoint, we conclude that X̃⫽X V,W . 䊏

IV. IRREDUCIBILITY AND UNITARY EQUIVALENCE

共1兲 The next two propositions decide when a triple of operators 兵 P,X V,W ,U 其 defined in the
preceding section is irreducible and when two such triples are unitarily equivalent. Here we shall
say that the triple 兵 P,X V,W ,U 其 on H is irreducible if any bounded operator A on H satisfying

PA傶A P, X V,W A傶AX V,W , and AU⫽UA 共14兲

is a scalar multiple of the identity operator on H.

Recall that the operator triple 兵 P,X V,W ,U 其 depends on the two families 兵 ␮ 1j,⫾ ; j苸I ⫾ 其 of
measures on the interval 关 q,1) and on the two unitary operators V,W:H⫺ ˜H⫹ . In order to
formulate the corresponding conditions it is convenient to work with the Hilbert spaces K⫾
⫽ 丣 j苸I ⫾ L 2 „关 q,1), ␮ 1j,⫾ … rather than with H⫾ ⫽ 丣 j苸I ⫾ L 2 „关 q,1),t ⫺1 ␮ 1j,⫾ …. Further, let P ⫾ denote
the self-adjoint operator on K⫾ which acts componentwise as the multiplication by the variable t.
Clearly, V and W are bounded linear operators of K⫺ to K⫹ such that
⫺1/2 ⫺1/2
V ⬘ª P ⫹
1/2
V P⫺ and W ⬘ª P ⫹
1/2
W P⫺ 共15兲

are unitary.
Proposition 2: The triple 兵 P,X V,W ,U 其 as defined above is irreducible if and only if any
bounded self-adjoint operators A ⫹ on K⫹ and A ⫺ on K⫺ satisfying

A ⫹ P ⫹ ⫽ P ⫹ A ⫹ , A ⫺ P ⫺ ⫽ P ⫺ A ⫺ , A ⫹ V ⬘ ⫽V ⬘ A ⫺ , A ⫹ W ⬘ ⫽W ⬘ A ⫺ 共16兲

Downloaded 14 Apr 2013 to 132.174.255.3. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jmp.aip.org/about/rights_and_permissions
 4602 J. Math. Phys., Vol. 40, No. 9, September 1999 Konrad Schmüdgen

or equivalently

A ⫹ P ⫹ ⫽ P ⫹ A ⫹ , A ⫺ P ⫺ ⫽ P ⫺ A ⫺ , A ⫹ V⫽VA ⫺ , A ⫹ W⫽WA ⫺ 共17兲

are scalar multiples of the identity.

Proof: We only show that the above condition implies the irreducibility of the triple. The
proof of the converse implication is easier and will be omitted. Suppose that A is a bounded
operator on H satisfying 共14兲. Since the set of such A is invariant under the involution, we can
assume that A is self-adjoint. Let E(•) denote the spectral projections of P. Since PA傶A P, the
subspace K⫹ ⫽E„关 q,1)…H of H reduces A and the restriction A ⫹ of A to K⫹ commutes with the
restriction P ⫹ of P to K⫹ . Similarly, the restrictions à ⫺ of A and P̃ ⫺ of P to the reducing
subspace E„(⫺1,q 兴 …H commute. Changing the variable from t to ⫺t, the Hilbert space
E„(⫺1,q 兴 …H and the operator P̃ ⫺ become K⫺ and P ⫺ , respectively, and the operator à ⫺ goes
into an operator, say A ⫺ , on K⫺ . Thus, A ⫺ P ⫺ ⫽ P ⫺ A ⫺ . From the assumptions AU⫽UA and
X V,W A債AX V,W it follows easily that (Af) ⫾ ⫾ ⫾ ⫾
e ⫽A ⫾ fe and (Af) o ⫽A ⫾ fo for f苸D(X V,W ). Since
Af苸D(X V,W ) has to satisfy the relation 共8兲, we obtain A ⫹ V⫽VA ⫺ and A ⫹ W⫽WA ⫺ . Therefore,
by the above condition, A ⫾ ⫽␭ ⫾ I for some ␭ ⫾ 苸C. Since A ⫹ V⫽VA ⫺ and AU⫽UA, it follows
that ␭ ⫹ ⫽␭ ⫺ and A⫽␭ ⫹ •I on H. 䊏
Using similar operator-theoretic arguments it is not difficult to prove the following.
Proposition 3: Two triples 兵 P,X V,W ,U 其 and 兵 P̃,X ˜V ,W
˜ ,Ũ 其 are unitarily equivalent if and only

if there are unitary operators A ⫹ of K⫹ to K̃⫹ and A ⫺ of K⫺ to K̃⫺ such that

A ⫹ P ⫹ ⫽ P̃ ⫹ A ⫹ , A ⫺ P ⫺ ⫽ P̃ ⫺ A ⫺ , A ⫹ V⫽ṼA ⫺ and A ⫹ W⫽W̃A ⫺ , 共18兲

where the tilde refers to the corresponding operators and spaces for the triple 兵 P̃,X ˜V ,W ˜ ,Ũ 其 .
共2兲 We shall illustrate the preceding by describing a few examples of irreducible representa-
tions. We begin with the simplest possible case.
Example 2: Suppose that the Hilbert spaces K⫹ and K⫺ are one-dimensional. Then the fami-
lies of measure 兵 ␮ ij,⫹ ;苸I ⫹ 其 and 兵 ␮ 1j,⫺ ; j苸I ⫺ 其 consist only of single Dirac measures ␦ a and ␦ b ,
respectively, where a,b苸 关 q,1). Then the triples 兵 P,X V,W ,U 其 are parametrized by complex num-
bers V⫽V ⬘ ⫽e i ␸ and W⫽W ⬘ ⫽e i ␺ , ␸ , ␺ 苸R. The self-adjoint extension X V,W is then characterized
by the boundary condition 共8兲, that is,

f⫹ ⫹ i␸ ⫺ ⫺
e ⫹fo ⫽e 共 fe ⫹fo 兲 , f⫹ ⫹ i␺ ⫺ ⫺
e ⫺fo ⫽e 共 fe ⫺fo 兲 .

Each such triple is irreducible because the condition in Proposition 2 is trivially fulfilled. Two
triples with different numbers VW ⫺1 are not unitarily equivalent. The case where e i ␸ ⫽e i ␺ ⫽1 and
a⫽b has been treated in detail in Ref. 3.
Example 3: Let P ⫹ be a self-adjoint operator and Z a unitary operator on a Hilbert space K⫹
such that the commutant 兵 P ⫹ ,Z 其 ⬘ is equal to C•I. Such operators exist on any separable Hilbert
space.10 Upon scaling we can assume that the spectrum of P ⫹ is contained in 关 q,1). By the
spectral representation theorem 共Ref. 11, chap. X, 5兲, we can represent P ⫹ up to unitary equiva-
lence as the multiplication operator by the independent variable t on some direct sum Hilbert space
K⫹ ⫽ 丣 j苸I ⫹ L 2 „关 q,1); ␮ 1j,⫹ …. Let 兵 ␮ 1j,⫺ ; j苸I 其 be an arbitrary family of measures on 关 q,1) such
that dim K⫹ ⫽dim K⫺ , where K⫺ ª 丣 j苸I ⫺ L 2 „关 q,1); ␮ 1j,⫺ …. Let W ⬘ be a unitary operator from K⫺
to K⫹ . We set V ⬘ ªZW ⬘ and define V and W by 共15兲. Then the triple 兵 P,X V,W ,U 其 is irreducible.
Indeed, if A ⫹ and A ⫺ be bounded self-adjoint operators satisfying 共17兲, then we have A ⫹ Z
⫽A ⫹ V ⬘ W ⬘ * ⫽V ⬘ A ⫺ W ⬘ * ⫽V ⬘ W ⬘ * A ⫹ ⫽ZA ⫹ and A ⫹ P ⫹ ⫽ P ⫹ A ⫹ , so that A ⫹ ⫽␭•I for some
␭苸C and hence A ⫺ ⫽V ⬘ * A ⫹ V ⬘ ⫽␭•I. By Proposition 2, the triple is irreducible. 䊏
Example 4: For this example we assume that there exist numbers a,b苸 关 q,1) such that
␮ 1j,⫹ ⫽ ␦ a and ␮ k,⫺
1 ⫽ ␦ b for all j苸I ⫹ and k苸I ⫺ . We shall show that in this case an irreducible
triple 兵 P,X V,W ,U 其 can be only obtained if both index sets I ⫹ and I ⫺ are singletons or equivalently
if dim K⫹ ⫽dim K⫺ ⫽1. Indeed, otherwise we take a self-adjoint operator A ⫹ on K⫹ such that

Downloaded 14 Apr 2013 to 132.174.255.3. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jmp.aip.org/about/rights_and_permissions
 J. Math. Phys., Vol. 40, No. 9, September 1999 Operator representations of a q-deformed . . . 4603

A ⫹ V ⬘ W ⬘ * ⫽V ⬘ W ⬘ * A ⫹ and A ⫹ 苸C•I and set A ⫺ ªV ⬘ * A ⫹ V ⬘ . Then the conditions 共16兲 are ful-
filled, hence the triple is not irreducible. 䊏
Example 5: If the spectra of the operators P ⫹ on K⫹ and P ⫺ on K⫺ are singletons, then we
have seen in Example 4 that irreducible triples exist only in the trivial case where I ⫹ and I ⫺ are
singletons. We now show that this is no longer true if both spectra consist of two points. To be
more precise, we shall consider the following situation: The index sets I ⫾ are disjoint unions of
two countable infinite sets I ⫾ 1 2
and I ⫾ and there are numbers a 1 ,a 2 ,b 1 ,b 2 苸 关 q,1), a 1 ⫽a 2 , such
that ␮ 1 ⫽ ␦ a 1 for j苸I ⫹ , ␮ 1 ⫽ ␦ a 2 for j苸I ⫹
j,⫹ 1 j,⫹ 2
, ␮ 1j,⫺ ⫽ ␦ b 1 for j苸I ⫺
1
, and ␮ 1j,⫺ ⫽ ␦ b 2 for j苸I ⫺
2
.
j
By identifying I ⫾ with the natural numbers the Hilbert spaces K⫹ and K⫺ become the direct sum
l 2 (N) 丣 l 2 (N) of two l 2 -spaces. We choose a bounded operator T on l 2 (N) such that 兵 T,T * 其 ⬘
⫽C•I and I⭐3T * T⭐2•I. It is well known 共see Ref. 12, Anhang, §4兲 that the operator matrix

Z⫽ 冉 T
⫺ 冑I⫺T * T
冑I⫺TT *
T*
冊
defines a unitary operator Z on K⫹ ⫽K⫺ ⫽l 2 (N) 丣 l 2 (N). Let W ⬘ be an arbitrary unitary operator
on K⫹ ⫽K⫺ and set V ⬘ ªZW ⬘ . Then the triple ( P,X V,W ,U) is irreducible.
Indeed, let A ⫹ and A ⫺ be self-adjoint bounded operators on K⫹ ⫽K⫺ satisfying 共17兲. Since
a 1 ⫽a 2 , the relation A ⫹ P ⫹ ⫽ P ⫹ A ⫹ implies that A ⫹ is given by a diagonal operator matrix

A ⫹⫽ 冉 冊
B
0 C
0
.

From 共17兲 we get A ⫹ Z⫽ZA ⫹ . Comparing the matrix entries of the first line yields BT⫽TB and
B 冑I⫺T * T⫽ 冑I⫺T * TC. Since B⫽B * , we have BT * ⫽T * B. Therefore, B commutes with T and
T * and so with 冑I⫺T * T which in turn gives 冑I⫺T * TB⫽ 冑I⫺T * TC. Because 冑I⫺T * T is
invertible, we get B⫽C. Since B苸 兵 T,T * 其 ⬘ , we obtain B⫽C⫽␭•I for some ␭苸C. Thus, A ⫹
⫽␭•I and A ⫺ ⫽V ⬘ * A ⫹ V ⬘ ⫽␭•I, so that the triple is irreducible by Proposition 2. 䊏

V. A CHARACTERIZATION OF THE OPERATOR TRIPLES

Let 兵 P,X V,W ,U 其 be an operator triple as in Sec. II and let D1 be the set of all vectors f⫽fX
⫹fe ⫹fo 苸D(X V,W ) with fX 苸Do , where Do is as defined in Sec. II. Then D1 is a dense linear
subspace of the Hilbert space H such that D1 is invariant under the operators P, X V,W , and U and
the restrictions of P and X V,W to D1 are essentially self-adjoint. Further, the three operators P,
X V,W , and U applied to vectors f苸D1 satisfy the relations 共1兲 and 共2兲. From the construction it is
clear that the range E„关 q,1)…H(⬵K⫹ ) of the spectral projection E„关 q,1)… of the operator P is
contained in D1 . Our next proposition says that the operator triples 兵 P,X V,W ,U 其 can be charac-
terized by some of the properties just mentioned.
Proposition 4: Let 兵 P ⬘ ,X ⬘ ,U ⬘ 其 be a triple of two self-adjoint operators P ⬘ and X ⬘ and a
unitary operator U ⬘ on a Hilbert space H⬘ . Let E(•) denote the spectral measure of P ⬘ . Suppose
that there exists a linear subspace D1 債D( P ⬘ X ⬘ )艚D(X ⬘ P ⬘ ) of H such that
共i兲 E„关 q,1)…H債D1 and E„(⫺1,⫺q 兴 …H債D1 .
共ii兲 The operators P ⬘ ,X ⬘ ,U ⬘ satisfy the relations (1) and (2) for vectors in D1 .
共iii兲 The restrictions P ⬘ d D1 and X ⬘ d D1 of P ⬘ and X ⬘ to D1 are essentially self-adjoint.
Then 兵 P ⬘ ,X ⬘ ,U ⬘ 其 is unitarily equivalent to an operator triple 兵 P,X V,W ,U 其 defined in
Sec. II.

Sketch of proof: The restriction P 1⬘ of P ⬘ to the invariant subspace H1 ªE„关 q,1)…H̃ is obvi-
ously a bounded self-adjoint operator on the Hilbert space H1 with spectrum contained in the
interval 关 q,1兴 . By the spectral representation theorem11, there is a family 兵 ␮ 1j,⫹ ; j苸I ⫹ 其 of finite
positive Borel measures on 关 q, 1 兴 and a unitary isomorphism of H1 on K⫹ ª 丣 j L 2 „关 q, 1 兴 , ␮ 1j,⫹ …

Downloaded 14 Apr 2013 to 132.174.255.3. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jmp.aip.org/about/rights_and_permissions
 4604 J. Math. Phys., Vol. 40, No. 9, September 1999 Konrad Schmüdgen

such that P 1⬘ is unitarily equivalent to the operator P 1 on K⫹ which acts componentwise as the
multiplication by the variable t. Since 1 is not an eigenvalue of P ⬘1 by construction, we have
␮ 1j,⫹ ( 兵 1 其 )⫽0 for all j苸I ⫹ . For simplicity let us identify H1 with K⫹ and P 1⬘ with P 1 .
Next we show that ker P ⬘ ⫽ 兵 0 其 . Let f苸ker P ⬘ . Since P ⬘ d D1 is essentially self-adjoint by 共iii兲,
there exists a sequence 兵 fn 其 of vectors fn 苸D1 such that fn ˜f and P ⬘ fn ˜ P ⬘ f⫽0 in H. Since
X ⬘ P ⬘ fn ⫽i(q 1/2U ⬘ * ⫹q 1/2U ⬘ )fn by 共ii兲 and the operators U ⬘ and U ⬘ * are bounded, we obtain
(q ⫺1/2U ⬘ * ⫹q 1/2U ⬘ )f⫽0 in the limit. This in turn yields that q 储 f储 ⫽ 储 f储 and so f⫽0.
By 共ii兲, we have U ⬘ P ⬘ f⫽q P ⬘ U ⬘ f for all f苸D1 . Since p ⬘ d D1 is essentially self-adjoint, this
remains valid for f苸D( P ⬘ ), so that P ⬘ 債qU ⬘ * P ⬘ U ⬘ . Since P ⬘ is self-adjoint, we conclude that
P ⬘ ⫽qU ⬘ * P ⬘ U ⬘ . Hence we have U ⬘ n E(N)⫽E(q ⫺n N) for any Borel subset N of R and arbitrary
n苸Z. Let ␮ j,⫹ be the extension of the measure ␮ 1j,⫹ to R⫹ as in Sec. II 共1兲. From the preceding
considerations it follows that E(R⫹ )H⫽ 丣 j L 2 (R⫹ , ␮ j,⫹ )⬅H⫹ and that U ⬘ acts in each compo-
nent by formula 共i兲 in Sec. II 共1兲. Proceeding in a similar manner, we obtain a family 兵 ␮ 1j,⫺ ; j
苸I ⫺ 其 of measures on 关 q,1兴 such that ␮ 1j,⫺ ( 兵 1 其 )⫽0 for j苸I ⫺ ,E(R⫺ )H⫽ 丣 j L 2 (R⫺ , ␮ j,⫺ )
⬅H⫺ in the notation of Sec. II and U ⬘ acts componentwise as given by formula 共i兲 in Sec. II 共1兲.
Since E( 兵 0 其 )H⫽ker P ⬘ ⫽ 兵 0 其 as proved in the preceding paragraph, we conclude that H⫽H⫹
丣 H⫺ .
From the construction it is clear that P ⬘ and U ⬘ are the operators P and U, respectively, as in
Sec. II. Let us finally turn to the operator X ⬘ . Recall that we have X ⬘ P ⬘ f⫽i(q ⫺1/2U ⬘ *
⫹q 1/2U ⬘ )f for f苸D1 . By arguing as in the paragraph before last, this relation remains valid for all
f苸D( P ⬘ ). If f denotes a component of the vector f, then the preceding equation yields that
gªt f 苸H, t ⫺1 g⫽ f 苸H and (X ⬘ g)(t)⫽i„q ⫺1 f (q ⫺1 t)⫺q f (qt)…⫽it ⫺1 „g(q ⫺1 t)⫺g(qt)…⫽(Xg)
⫻(t). Hence X ⬘ f⫽Xf for all f苸D( P ⬘ ). Since X ⬘ d D1 is essentially self-adjoint, the relation
U ⬘ X ⬘ f⫽q ⫺1 X ⬘ U ⬘ f for f苸D1 by 共ii兲 extends to vectors f苸D(X ⬘ ), so that U ⬘ X ⬘ U ⬘ * ⫽q ⫺1 X ⬘ .
Thus, X ⬘ is a self-adjoint extension of the operator X such that UD(X ⬘ )⫽D(X ⬘ ). By Proposition
1, X ⬘ is of the form X V,W . 䊏

VI. *-REPRESENTATIONS OF THE q-HEISENBERG ALGEBRA

共1兲 We have considered so far only operator triples and operator relations rather than repre-
sentations of the algebra A(q). However, any operator triple 兵 P,X V,W ,U 其 gives rise to a self-
adjoint representation of the *-algebra as follows. Indeed, let D1 be the domain defined at the
beginning of Sec. IV. For vectors in D1 the operators P,X V,W ,U satisfy the defining relations 共1兲
and 共2兲 of the algebra A(q). Hence there exists a unique *-representation ␲ 1 of the *-algebra
A(q) on the domain D1 such that

␲ 1 共 p 兲 ⫽ P d D1 , ␲ 1 共 x 兲 ⫽X V,W d D1 , ␲ 1 共 u 兲 ⫽U d D1 .

共For the notions on unbounded *-representations used in what follows we refer to Ref. 13. Recall
that the symbol T d D1 means the restriction of T to D1 .兲
The *-representation ␲ 1 is not yet self-adjoint 共see Ref. 13, Definition 8.1.10兲, because,
roughly speaking, D1 is not the largest possible domain. However, since the operators ␲ 1 (p) and
␲ 1 (x) are essentially self-adjoint, it follows at once from Proposition 8.1.12共v兲 in Ref. 13 that the
adjoint representation ␲ ª( ␲ 1 ) * is self-adjoint. It is not difficult to verify that the domain D of
the *-representation ␲ is just the intersection of domains of all possible products of the operators
P,X V,W ,U 共see Ref. 13, Proposition 8.1.17兲. From these facts it follows that the operator triple
兵 P,X V,W ,U 其 is irreducible if and only if the *-representation ␲ is so and that two triples are
unitarily equivalent if and only if the corresponding *-representations are so. That is, Propositions
2 and 3 provide also the conditions for the irreducibility and the unitary equivalence of these
*-representations of the *-algebra A(q).
共2兲 Finally, we briefly discuss how operator representations of the q-deformed Heisenberg
algebra A(q) can be constructed by means of the Schrödinger representation Pª⫺i(d/dt) and
Qªt of the ‘‘ordinary’’ momentum and position operators.

Downloaded 14 Apr 2013 to 132.174.255.3. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jmp.aip.org/about/rights_and_permissions
 J. Math. Phys., Vol. 40, No. 9, September 1999 Operator representations of a q-deformed . . . 4605

Let us write q⫽e ⫺ ␣ with ␣ 苸R. We define three operators U, P, and X on the Hilbert space
H⫽L 2 (R):

U⫽e i Q , P⫽e ␣ P, X⫽i共 q ⫺1/2e ⫺i Q⫺q 1/2e i Q兲 e ⫺ ␣ P. 共19兲

2
The vector space DªLin兵 e ␥ t⫺t ; ␥ 苸C其 is a dense linear subspace of H. Since the operator
e ␤ P, ␤ 苸R, acts as (e ␤ P f )(t)⫽ f (t⫺ ␤ i) on functions f 苸C 共see, for instance, Ref. 14 for a
rigorous proof兲, the operators U, P, and X satisfy the relations 共1兲 and 共2兲⬘ and the restrictions of
these operators to the invariant dense domain D define a *-representation of the *-algebra A(q).
This operator representation 共19兲 appears already somewhat hidden in Ref. 1. Indeed, if we change
the variable t to e t , then the operator triple 兵 U 丣 U,(⫺ P) 丣 P,(⫺X) 丣 X 其 on the direct sum Hilbert
space H 丣 H is easily seen to be unitarily equivalent to the triple in formula 共2.2兲 in Ref. 1.
The operator representation 共19兲 is irreducible on H. Obviously, U is unitary and P is self-
adjoint. However, an essential disadvantage of the representation 共19兲 is that the operator X is only
symmetric, but not essentially self-adjoint. The latter can be shown by the argument used in the
proof of Proposition A.2 in Ref. 14. The reason for this failure is the fact the holomorphic function
h(z)⫽q ⫺1/2e i z ⫺q 1/2e ⫺i z admits the zero z o ⫽i( ␣ /2) in the strip 兵 z苸C: 0⬍Im z⬍␣其.

1
M. Fichtmüller, A. Lorek, and J. Wess, ‘‘q-Deformed phase space and its lattice structure,’’ Z. Phys. C 71, 533–538
共1996兲.
2
A. Hebecker, S. Schreckenberg, J. Schwenk, W. Weich, and J. Wess, ‘‘Representations of a q-deformed Heisenberg
algebra,’’ Z. Phys. C 64, 335–359 共1994兲.
3
A. Lorek, W. Weich, and J. Wess, ‘‘Non-commutative Euclidean and Minkowski structures,’’ preprint, MPI-PhI/96-124,
Munich 共1996兲.
4
J. Schwenk and J. Wess, ‘‘A q-deformed quantum mechanical toy model,’’ Phys. Lett. B 291, 273–277 共1992兲.
5
W. Weich, ‘‘The Hilbert space representations for SO q (3)-symmetric quantum mechanics,’’ preprint, Munich,
hep-th/9404029 共1994兲.
6
J. Wess and B. Zumino, ‘‘Covariant differential calculus on the quantum hyperplane,’’ Nucl. Phys. B, Proc. Suppl. 18B,
302–312 共1991兲.
7
W. Pusz and S. L. Woronowicz, ‘‘Twisted second quantization,’’ Rep. Math. Phys. 27, 231–257 共1989兲.
8
T. H. Koornwinder and R. F. Swartouw, ‘‘On q-analogues of the Fourier and Hankel transforms,’’ Trans. Am. Math.
Soc. 333, 445–461 共1992兲.
9
H. T. Koelink and R. F. Swartouw, ‘‘On the zeros of the Hahn-Exton q-Bessel function and associated q-Lommel
polynomials,’’ J. Math. Anal. Appl. 186, 690–770 共1994兲.
10
D. M. Topping, Lectures on von Neumann Algebras 共Van Nostrand, New York, 1971兲.
11
N. Dunford and J. Schwartz, Linear Operators, Part II 共Interscience, New York, 1963兲.
12
F. Riesz and B. Sz.-Nagy, Vorlesungen überFunktionalanalysis 共DVW, Berlin, 1956兲.
13
K. Schmüdgen, Unbounded Operator Algebras and Representation Theory 共Birkhäuser, Basel, 1990兲.
14
K. Schmüdgen, ‘‘Integrable operator representations of R2q , X q, ␥ and SL q (2,R),’’ Commun. Math. Phys. 159, 217–237
共1992兲.

Downloaded 14 Apr 2013 to 132.174.255.3. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jmp.aip.org/about/rights_and_permissions