Abstract
We study some complete orthonormal systems on the real line. These systems are determined by Bargmann-type transforms, which are Fourier integral operators with complex-valued quadratic phase functions. Each system consists of eigenfunctions for a second-order elliptic differential operator like the Hamiltonian of the harmonic oscillator. We also study the commutative case of a certain class of systems of second-order differential operators called the non-commutative harmonic oscillators. By using the diagonalization technique, we compute the eigenvalues and eigenfunctions for the commutative case of the non-commutative harmonic oscillators. Finally, we study a family of functions associated with an ellipse in the phase plane. We show that the family is a complete orthogonal system on the real line.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
We are concerned with the Bargmann-type transform defined by
where \(\phi (z,x)\) is a complex-valued quadratic phase function of the form
with assumptions \(B\ne 0\) and \({\text {Im}}\,C>0\), and \(C_\phi =2^{-1/2}\pi ^{-3/4}|{B}|({\text {Im}}\,C)^{-1/4}\). Throughout the present paper, we deal with only the one-dimensional case for the sake of simplicity. It is possible to discuss higher-dimensional case, and we omit the detail. Note that the integral transform (1) is well defined for tempered distributions on \({\mathbb {R}}\) since \({\text {Re}}(i\phi (z,x))={\mathcal {O}}(-{\text {Im}}\, C x^2/3)\) for \(|{x}|\rightarrow \infty \).
These integral transforms were introduced by Sjöstrand (See, e.g., [13]). He developed microlocal analysis based on them. One can see (1) as a global Fourier integral operator associated with a linear canonical transform \(\kappa _T:{\mathbb {C}}^2\ni (x,-\phi ^\prime _x(z,x))\mapsto (z,\phi ^\prime _z(z,x))\in {\mathbb {C}}^2\), that is,
If we set \(\Phi (z)=\max _{x\in {\mathbb {R}}}{\text {Re}}(i\phi (z,x))\), then we have
This means that the singularities of a tempered distribution u described in the phase plane \({\mathbb {R}}^2\) are translated into those of \(T_hu\) in the I-Lagrangian submanifold \(\kappa _T({\mathbb {R}}^2)\). The microlocal analysis of Sjöstrand is based on the equivalence of the Weyl quantization on \({\mathbb {R}}\), the Weyl quantization on \(\kappa _T({\mathbb {R}})\), and the Berezin–Toeplitz quantization on \({\mathbb {C}}\). For more detail about them, see [13] or [4].
Let \(L^2({\mathbb {R}})\) be the set of all square-integrable functions on \({\mathbb {R}}\), and let \(L^2_\Phi ({\mathbb {C}})\) be the set of all square-integrable functions on \({\mathbb {C}}\) with respect to a weighted measure \(e^{-2\Phi (z)/h}L(\mathrm{d}z)\), where L is the Lebesgue measure on \({\mathbb {C}}\simeq {\mathbb {R}}^2\). We denote by \({\mathscr {H}}_\Phi ({\mathbb {C}})\) the set of all entire functions in \(L^2_\Phi ({\mathbb {C}})\). It is well known that \(T_h\) gives a Hilbert space isomorphism of \(L^2({\mathbb {R}})\) onto \({\mathscr {H}}_\Phi ({\mathbb {C}})\), that is,
where
We sometimes denote \((U,V)_{L^2_\Phi }\) for \(U,V \in {\mathscr {H}}_\Phi ({\mathbb {C}})\) by \((U,V)_{{\mathscr {H}}_\Phi }\). The inverse mapping \(T_h^*\) is given by
Note that \(T_h^*\) is well defined for \(U \in L^2_\Phi ({\mathbb {C}})\). \(T_h{\circ }T_h^*\) becomes an orthogonal projector of \(L^2_\Phi ({\mathbb {C}})\) onto \({\mathscr {H}}_\Phi ({\mathbb {C}})\). More concretely,
where \(C_\Phi =|{B}|^2/(2\pi \,{\text {Im}}\,C)\), and \(\Psi (z,\zeta )\) is a holomorphic quadratic function defined by the critical value of \(-\{\phi (z,X)-\overline{\phi ({\bar{\zeta }},{\bar{X}})}\}/2i\) for \(X\in {\mathbb {C}}\), that is,
In particular, \(U=T_h{\circ }T_h^*U\) for \(U \in {\mathscr {H}}_\Phi ({\mathbb {C}})\), and \({\mathscr {H}}_\Phi ({\mathbb {C}})\) becomes a reproducing kernel Hilbert space.
Here, we recall elementary facts related with the classical Bargmann transform \(B_h\) which is the most important example of \(T_h\). This was introduced by Bargmann in [3]. We can also refer [8] for this. The Bargmann transform \(B_h\) on \({\mathbb {R}}\) is defined by
Note that the integral kernel of \(B_h\) is the generating function of Hermite functions. We denote \(L_\Phi ({\mathbb {C}})\) and \({\mathscr {H}}_\Phi ({\mathbb {C}})\) for \(B_h\) by \(L^2_B({\mathbb {C}})\) and \({\mathscr {H}}_B({\mathbb {C}})\), respectively, those are,
and \({\mathscr {H}}_B({\mathbb {C}})=\{U(z) \in L^2_B({\mathbb {C}})\ \vert \ \partial U/\partial {\bar{z}}=0\}\). The Bargmann projector, which is the orthogonal projection of \(L_B^2({\mathbb {C}})\) onto \({\mathscr {H}}_B({\mathbb {C}})\), is given by
In view of the Taylor expansion of the reproducing kernel \(e^{z{\bar{\zeta }}/2h}/(2\pi h)\), the formula \(U=B_h{\circ }B_h^*U\) for \(U \in {\mathscr {H}}_B({\mathbb {C}})\) becomes
A family of functions \(\{\varphi _{B,n}\}_{n=0}^\infty \) is a complete orthonormal system of \({\mathscr {H}}_B({\mathbb {C}})\) since \({\mathscr {H}}_B({\mathbb {C}})\) is the set of all entire functions belonging to \(L^2_B({\mathbb {C}})\).
We shall see more detail about \(\{\varphi _{B,n}\}_{n=0}^\infty \). We set for \(U \in {\mathscr {H}}_B({\mathbb {C}})\)
Actually, \(\Lambda _B^*\) is the adjoint of \(\Lambda _B\) on \({\mathscr {H}}_B({\mathbb {C}})\). Elementary computation gives
We shall pullback these facts on \({\mathbb {R}}\) by using \(B_h^*\). Set
\(\phi _{B,n}\) is said to be the nth Hermite function, and a family \(\{\phi _{B,n}\}_{n=0}^\infty \) is a complete orthonormal system of \(L^2({\mathbb {R}})\) since \(B_h\) is a Hilbert space isomorphism of \(L^2({\mathbb {R}})\) onto \({\mathscr {H}}_B({\mathbb {C}})\). Operators
are said to be annihilation and creation operators, respectively. Note that
Then, we have so-called Rodrigues formula
Set \(H_B=P_B^*\circ P_B + h = P_B \circ P_B^*- h\). Then
which is said to be the Hamiltonian of the harmonic oscillator. Equation (6) becomes
Thus, the nth Hermite function \(\phi _{B,n}\) is an eigenfunction of \(H_B\) for the nth eigenvalue \((2n+1)h\).
The purpose of the present paper is to study the generalization of the known facts on the usual Bargmann transform \(B_h\). The plan of this paper is as follows. In Sect. 2, we study the general Bargmann-type transform (1) and obtain generalized annihilation and creation operators, the Hamiltonian of the generalized harmonic oscillator and its eigenvalues, generalized Hermite functions and the Rodrigues formula. In Sect. 3, we study a \(2\times 2\)-system of second-order ordinary differential operators, which is said to be a non-commutative harmonic oscillators. More precisely, we study the commutative case of the non-commutative harmonic oscillators and obtain the eigenvalues and eigenfunctions by using our original elementary computation. Finally, in Sect. 4, we study the general Bargmann-type transform (1) which might be related with ellipses in the phase plane \({\mathbb {R}}^2\).
2 Modified Harmonic Oscillators and Hermite Functions
In this section, we study the general form of the Bargmann-type transform (1). We remark that the choice of the constant A in the phase function is not essential. We can choose
Then, (3) and (5) become very simple as
respectively. Moreover, the orthogonal projector (4) of \(L^2_\Phi ({\mathbb {C}})\) onto \({\mathscr {H}}_\Phi ({\mathbb {C}})\), and the I-Lagrangian submanifold (2) become
respectively. Recall \(U(z)=T_h \circ T_h^*U(z)\) for all \(U \in {\mathscr {H}}_\Phi ({\mathbb {C}})\). If we consider the Taylor expansion of \(e^{-|{Bz}|^2/2h \,{\text {Im}}\, C}\), we have for \(U \in {\mathscr {H}}_\Phi ({\mathbb {C}})\),
Theorem 2.1
The family of monomials \(\{\varphi _n\}_{n=0}^\infty \) is a complete orthonormal system of \({\mathscr {H}}_\Phi ({\mathbb {C}})\).
Proof
The completeness is obvious. We have only to show that \((\varphi _m,\varphi _n)_{{\mathscr {H}}_\Phi }=\delta _{mn}\), where \(\delta _{mn}\) is Kronecker’s delta. Without loss of generality, we may assume that \(m \geqslant n\). By using the integration by parts and the change of variable \(\zeta =Bz/\sqrt{h \,{\text {Im}}\, C}\), we deduce that
This completes the proof. \(\square \)
Set \(\phi _n(x)=T_h^*\varphi _n(x)\), \(n=0,1,2,\ldots \). Since \(T_h\) is a Hilbert space isomorphism of \(L^2({\mathbb {R}})\) onto \({\mathscr {H}}_\Phi ({\mathbb {C}})\), we have the following.
Theorem 2.2
\(\{\phi _n\}_{n=0}^\infty \) is a complete orthonormal system of \(L^2({\mathbb {R}})\).
In what follows we study the family of functions \(\{\phi _n\}_{n=0}^\infty \) in detail. Let \(\Lambda \) be a linear operator on \({\mathscr {H}}_\Phi ({\mathbb {C}})\) defined by
Its Hilbert adjoint is
We call \(\Lambda \) and \(\Lambda ^*\) annihilation and creation operators on \({\mathscr {H}}_\Phi ({\mathbb {C}})\), respectively. Since \(\varphi _n\) is a monomial of degree n, we have for \(n=0,1,2,\ldots \)
We shall pullback these facts by using \(T_h^*\). Set
To state the concrete form of H, we introduce the Weyl pseudodifferential operators. For an appropriate function \(a(x,\xi )\) of \((x,\xi )\in {\mathbb {R}}^2\), its Weyl quantization is defined by
for \(u \in {\mathscr {S}}({\mathbb {R}})\), where \({\mathscr {S}}({\mathbb {R}})\) denotes the Schwartz class on \({\mathbb {R}}\). Set \(D_x=-i\mathrm{d}/\mathrm{d}x\) for short.
Here, we give the concrete forms of operators P, \(P^*\) and H on \({\mathbb {R}}\).
Proposition 2.3
We have
Proof
We first compute P and \(P^*\). Since \(\Lambda \circ T_h = T_h \circ P\), we deduce that for any \(u \in {\mathscr {S}}({\mathbb {R}})\),
which shows that \(P=-(hD_x+{\bar{C}}x)/{\bar{B}}\). In the same way, we can obtain \(P^*=-(hD_x+Cx)/B\), which is certainly the adjoint of P on \(L^2({\mathbb {R}})\). Next, we compute H. Simple computation gives
which completes the proof. \(\square \)
By using the pullback of (7) and (8), we have for \(n=0,1,2,\ldots \)
If we compute the concrete form of \(\phi _0\), then we obtain the Rodrigues formula for \(\{\phi _n\}_{n=0}^\infty \).
Theorem 2.4
We have for \(n=0,1,2,\ldots \)
Proof
Recall the definition of \(\phi _0\). We have
where
Change the variable \(\zeta =\xi +i\eta :={\overline{Bz}}\), \((\xi ,\eta )\in {\mathbb {R}}^2\). We deduce
Then, we can obtain
This completes the proof. \(\square \)
3 The Commutative Case of Non-commutative Harmonic Oscillators
Consider a \(2\times 2\) system of second-order differential operators of the form
where \(\alpha \) and \(\beta \) are positive constants satisfying \(\alpha \beta >1\), and
A matrix \(A_{(\alpha ,\beta )}(\xi ^2+x^2)/2+J(ix\xi )\), which is the symbol of the operator \(Q_{(\alpha ,\beta )}\), is a Hermitian matrix, and all its eigenvalues are real valued. Note that all its eigenvalues are positive for \((x,\xi )\ne (0,0)\) if and only if \(\alpha \beta >1\). In other words, \(Q_{(\alpha ,\beta )}\) is a system of semiclassical elliptic differential operators if and only if \(\alpha \beta >1\). The system of differential operators \(Q_{(\alpha ,\beta )}\) was mathematically introduced in [10] by Parmeggiani and Wakayama. They call \(Q_{(\alpha ,\beta )}\) a Hamiltonian of non-commutative harmonic oscillator. The word “non-commutative” comes from the non-commutativity \(A_{(\alpha ,\beta )}J \ne JA_{(\alpha ,\beta )}\) for \(\alpha \ne \beta \). It is not known that the system of differential equations for \(Q_{(\alpha ,\beta )}\) describes a physical phenomenon.
Parmeggiani and Wakayama intensively studied spectral properties of \(Q_{(\alpha ,\beta )}\) in [10,11,12]. See also a monograph [9]. They proved that if \(\alpha \beta >1\), then \(Q_{(\alpha ,\beta )}\) is a self-adjoint and positive operator, and its spectra consists of positive eigenvalues whose multiplicities are at most three. In case of \(\alpha =\beta \), they obtained more detail.
The purpose of the present section is to give alternative proof of the results of Parmeggiani and Wakayama for the commutative case \(\alpha =\beta \). More precisely, we study \(Q_{(\alpha ,\alpha )}\) by using the results in the previous section.
In what follows we assume that \(\alpha =\beta \). Then, \(\alpha >1\) since \(\alpha >0\) and \(\alpha ^2=\alpha \beta >1\). Let I be the \(2\times 2\) identity matrix. Set \(Q_\alpha =Q_{(\alpha ,\alpha )}\) for short. Let U be a \(2\times 2\) unitary matrix defined by
which diagonalize J as
Then, we have
Note that \(|\nu _{\alpha ,\pm }|=1\) and \({\text {Im}}\,\nu _{\alpha ,\pm }>0\).
Here, we make use of the results in the previous section by setting
Note that the requirement \({\text {Im}}\,C>0\) is satisfied. Set
for \(n=0,1,2,\ldots \). Then, we deduce that \(\{\phi _{\alpha ,\pm ,n}\}_{n=0}^\infty \) is a complete orthonormal system of \(L^2({\mathbb {R}})\), and
In order to get the eigenfunctions of \(Q_\alpha \), we set
those are,
We have proved the results of this section as follows.
Theorem 3.1
A system of \({\mathbb {C}}^2\)-valued functions \(\bigl \{\Phi _{\alpha ,\mu ,n} \ \big \vert \ \mu =\pm , n=0,1,2,\ldots .\bigr \}\) is a complete orthonormal system of \(L^2({\mathbb {R}}) \oplus L^2({\mathbb {R}})\), and satisfies
This is not a new result. This was first proved by Parmeggiani and Wakayama in [10]. We believe that our method of proof is easier than that of [10].
4 Orthogonal Systems Associated with Ellipses in the Phase Plane
Throughout of the present section, we assume that \(h=1\) for the sake of simplicity. We begin with recalling the relationship between the standard Bargmann transform \(B_1\) and circles in the phase plane. Here, we introduce a Berezin–Toeplitz quantization on \({\mathbb {C}}\). Let b(z) be an appropriate function on \({\mathbb {C}}\). Set
It is known that
for \(u,v \in {\mathscr {S}}({\mathbb {R}})\). See, e.g., [13] and [4]. The operator \({\tilde{T}}_b\) is said to be the Berezin–Toeplitz quantization of b, which acts on \({\mathscr {H}}_B({\mathbb {C}})\). If b is a characteristic function on \({\mathbb {C}}\), then \({\text {Op}}^\text {W}_1(a)\) is said to be a Daubechies’ localization operator introduced in [7]. Moreover, Daubechies proved that if b(z) is radially symmetric, that is, b is of the form \(b(x-i\xi )=c(x^2+\xi ^2)\) with some function c(s) for \(s\geqslant 0\), then all the usual Hermite functions \(\phi _{B,n}\) (\(n=0,1,2,\ldots \)) are the eigenfunctions of \({\tilde{T}}_b\):
Recently, Daubechies’ results have been developed. Here, we quote two interesting results of inverse problems studied in [1] and [15]. On one hand, in [15] Yoshino proved that radially symmetric symbols of the Berezin–Toeplitz quantization on \({\mathbb {C}}\) can be reconstructed by all the eigenvalues \(\{\lambda _n\}_{n=0}^\infty \). More precisely, by using the framework of hyperfunctions, he obtained the reconstruction formula for radially symmetric symbols. On the other hand, in [1] Abreu and Dörfler studied the inverse problem for Daubechies’ localization operators. Let \(\Omega \) be a bounded subset of \({\mathbb {C}}\), and let b(z) be the characteristic function of \(\Omega \). They proved that if there exists a nonnegative integer m such that the mth Hermite function \(\phi _{B,m}\) is an eigenfunction of \({\text {Op}}^\text {W}_1(a)\), then \(\Omega \) must be a disk centered at the origin. In this case, it follows automatically that all the Hermite functions \(\phi _{B,n}\) are eigenfunctions of \({\text {Op}}^\text {W}_1(a)\) associated with eigenvalues
respectively, where R is the radius of \(\Omega \). In particular \(R=-\log (1-\lambda _0)\). That is the review of the relationship between the usual Bargmann transform and circles (or disks) in \({\mathbb {C}}\).
The purpose of the present section is to consider the possibility of the extension of the above to ellipses (or elliptic disks) in \({\mathbb {C}}\). Unfortunately, however, we could not obtain the extension of the above. In what follows we introduce a family of functions which might be concerned with ellipses in \({\mathbb {C}}\). Here, it is worth to mention the interesting work [14] of van Eijndhoven and Meyers. They introduced for \(0<s<1\) function spaces \(\chi _s({\mathbb {C}})\), which is the set of all entire functions \(\varphi (z)\) on \({\mathbb {C}}\) satisfying
As the author pointed out in [5], \(\chi _s({\mathbb {C}})\) is determined by the ellipse on \({\mathbb {C}}\) of the form
and \(\chi _s({\mathbb {C}})\) is a special case of \({\mathscr {H}}_\Phi ({\mathbb {C}})\) with
Recently, Ali, Górska, Horzela and Szafraniec in [2] studied some kinds of generating functions of Hermite polynomials in the abstract setting and introduced some ortonormalized holomorphic Hermite functions in some function spaces including \(\chi _s({\mathbb {C}})\). Here, we introduce a holomorphic Hermite functions \(\{h_n\}_{n=0}^\infty \) on \({\mathbb {C}}\) and normalizing constants b(n) defined by
One of the interesting results of [2] is that \(\{h_n/\sqrt{b(n)}\}_{n=0}^\infty \) is a complete orthonormal system of \(\chi _s({\mathbb {C}})\). See [6] for more information on general holomorphic Hermite functions and their basic properties.
Let \(\alpha >0\) and let \(\beta \in {\mathbb {R}}\). Suppose that \((\alpha ,\beta )\ne (1,0)\). For \(\rho >0\),
is an elliptic disk in \({\mathbb {C}}\). Note that \(E_\rho \) is a usual disk if and only if \((\alpha ,\beta )=(1,0)\). Note that
is the set of all ellipses centered at the origin, where \(\partial {E_\rho }=\{x-i\xi \in {\mathbb {C}}\ \vert \ |\alpha {x}-i(\beta {x}+\xi )|= \rho \}\). Indeed, consider a function
Elementary computation gives
Here, we introduce a function \(\psi _0(z)\) which seems to be related with an elliptic disk \(E_\rho \). Set \(z=x-i\xi \) and \(\zeta =\alpha {x}-i(\beta {x}+\xi )\) for \((x,\xi )\in {\mathbb {R}}^2\), \(\alpha >0\) and \(\beta \in {\mathbb {R}}\). Then,
We define the function \(\psi _0(z)\) by
Let \(||\cdot ||_{{\mathscr {H}}_B}\) be the norm of \({\mathscr {H}}_B({\mathbb {C}})\) determined by the inner product \((\cdot ,\cdot )_{{\mathscr {H}}_B}\). The properties of \(\psi _0(z)\) are the following.
Lemma 4.1
We have
- (i)
\(\psi _0\in {\mathscr {H}}_B({\mathbb {C}})\).
- (ii)
\(|\psi _0(z)|^2e^{-|{z}|^2/2}=e^{-|\zeta |^2/(\alpha ^2+\beta ^2+1)}\).
- (iii)
\(||\psi _0||_{{\mathscr {H}}_B}^2=(\alpha ^2+\beta ^2+1)\pi /\alpha \).
Proof
We first show (i). We have only to show the integrability of \(|\psi _0(z)|^2e^{-|{z}|^2/2}\) since \(\psi _0(z)\) is an entire function. Note that
We have
since
for \((\alpha ,\beta )\ne (1,0)\). Thus, \(0<|{a}|<1\). We deduce that
This implies that \(|\psi _0(z)|^2e^{-|{z}|^2/2}\) is integrable on \({\mathbb {C}}\) with respect to the Lebesgue measure \(L(\mathrm{d}z)\) and \(\psi _0\in {\mathscr {H}}_B({\mathbb {C}})\).
We show (ii) and (iii). Elementary computation shows that
which implies (ii). Moreover, it is easy to see that \(\mathrm{d}z \wedge \mathrm{d}{\bar{z}}=\alpha ^{-1}\mathrm{d}\zeta \wedge \mathrm{d}{\bar{\zeta }}\) and
This completes the proof. \(\square \)
The identity \(|\psi _0(z)|^2e^{-|{z}|^2/2}=e^{-|\zeta |^2/(\alpha ^2+\beta ^2+1)}\) makes us to expect that \(\psi _0\) might be related with an elliptic disk \(E_\rho \) and generate a family of eigenfunctions for the Daubechies’ localization operators supported in \(E_\rho \). Unfortunately, however, this expectation fails to hold. The purpose of the present section is to generate a family of functions by \(\psi _0\) and show its properties similar to the previous sections. To state our results in the present section, we here introduce some notation. Set
It is easy to see that \(a+2\lambda =1/{\bar{a}}\), \(\Lambda _{\alpha ,\beta }\psi _0=0\), and \(\Lambda _{\alpha ,\beta }^*\) is the Hilbert adjoint of \(\Lambda _{\alpha ,\beta }\) on \({\mathscr {H}}_B({\mathbb {C}})\). We make use of \(\psi _0\), \(\Lambda _{\alpha ,\beta }\) and \(\Lambda _{\alpha ,\beta }^*\) as a generating element of a family of functions, and annihilation and creation operators, respectively. Set \(\psi _n=(\Lambda _{\alpha ,\beta }^*)^n\psi _0\) for \(n=0,1,2,\ldots .\), and set
for short. Properties of \(\Lambda _{\alpha ,\beta }\), \(\Lambda _{\alpha ,\beta }^*\) and \(\{\psi _n\}_{n=0}^\infty \) are the following.
Theorem 4.2
-
(i)
\(\{\psi _n\}_{n=0}^\infty \) satisfies a formula of the form
$$\begin{aligned} \psi _n(z) = \left\{ e^{-\lambda z^2/2} \left( \frac{\partial }{\partial z}\right) ^n e^{\lambda z^2/2} \right\} \psi _0(z), \quad n=0,1,2,\ldots . \end{aligned}$$ -
(ii)
For \(n=1,2,3,\ldots \),
$$\begin{aligned} \Lambda _{\alpha ,\beta } (\Lambda _{\alpha ,\beta }^*)^n = (\Lambda _{\alpha ,\beta }^*)^n \Lambda _{\alpha ,\beta } + n \frac{\lambda }{a} (\Lambda _{\alpha ,\beta }^*)^{n-1}. \end{aligned}$$ -
(iii)
\(\{\psi _n\}_{n=0}^\infty \) is a complete orthogonal system of \({\mathscr {H}}_B({\mathbb {C}})\).
-
(iv)
For \(n=0,1,2,\ldots .\),
$$\begin{aligned} \left( \frac{\Lambda _{\alpha ,\beta }^*\Lambda _{\alpha ,\beta }}{|{C_{\alpha ,\beta }}|^2} + \frac{\alpha ^2}{1+\beta ^2} \right) \psi _n = \frac{\alpha ^2}{1+\beta ^2} (2n+1) \psi _n. \end{aligned}$$
Proof
First we show (i). Note that for any \(c\in {\mathbb {C}}\) and for any holomorphic function f(z), we deduce that
Using this repeatedly, we have
which is desired.
Next we show (ii). We employ induction on n. For \(n=1\), we deduce that
Here, we suppose that (ii) holds for some \(n=1,2,3,\ldots .\). We show the case of \(n+1\). By using the cases of n and 1, we deduce that
which is desired.
For (iii), we here show only the orthogonality
The completeness will be automatically proved later. Recall that \(\Lambda _{\alpha ,\beta }\psi _0=0\). Suppose that \(m \geqslant n\). By using (ii) repeatedly, we deduce that
If \(m>n\), then
If \(m=n\), then
Finally, we show (iv). By using (ii) and \(\Lambda _{\alpha ,\beta }\psi _0=0\) again, we deduce that
This completes the proof. \(\square \)
Here, we introduce a family of functions \(\{\Psi _n\}_{n=0}^\infty \) on \({\mathbb {R}}\) by setting \(\Psi _n=B_1^*\psi _n\) for \(n=0,1,2,\ldots \). In order to study basic properties on \(\{\Psi _n\}_{n=0}^\infty \), we introduce notation. Set
for short. Then, we have
Set
where we take its argument in \((-\pi /4,\pi /4)\). Our results in the present section are the following.
Theorem 4.3
-
(i)
We have for \(n=0,1,2,\ldots .\)
$$\begin{aligned} \Psi _n(x) =&A_{\alpha ,\beta } (-C_{\alpha ,\beta })^n \exp \left( -i \frac{(\alpha ^2+\beta ^2+1)\beta }{2(1+\beta ^2)} x^2 \right) \\&\times \exp \left( \frac{\alpha ^2}{2(1+\beta ^2)} x^2 \right) \left( \frac{\mathrm{d}}{\mathrm{d}x}\right) ^n \exp \left( - \frac{\alpha ^2}{1+\beta ^2} x^2 \right) . \end{aligned}$$In particular,
$$\begin{aligned} \Psi _0(x) = A_{\alpha ,\beta } \exp \left( - \frac{\alpha ^2+i(\alpha ^2+\beta ^2+1)\beta }{2(1+\beta ^2)} x^2 \right) . \end{aligned}$$ -
(ii)
\(\{\Psi _n\}_{n=0}^\infty \) is a family of eigenfunctions of \(H_{\alpha ,\beta }\), that is,
$$\begin{aligned} H_{\alpha ,\beta }\Psi _n = \frac{\alpha ^2}{1+\beta ^2} (2n+1) \Psi _n, \quad n=0,1,2,\ldots .. \end{aligned}$$ -
(iii)
\(\{\Psi _n\}_{n=0}^\infty \) is a complete orthogonal system of \(L^2({\mathbb {R}})\).
Recall that Theorem 4.2 was proved except for the completeness of \(\{\psi _n\}_{n=0}^\infty \). Theorem 4.2 without the completeness implies (ii) of Theorem 4.3 and the orthogonality of \(\{\Psi _n\}_{n=0}^\infty \) in \(L^2({\mathbb {R}})\). If (i) of Theorem 4.3 holds, then the completeness of \(\{\Psi _n\}_{n=0}^\infty \) in \(L^2({\mathbb {R}})\) follows immediately. Indeed combining (i) of Theorem 4.3 and the results in Sect. 2 with
we can check the completeness of \(\{\Psi _n\}_{n=0}^\infty \) in \(L^2({\mathbb {R}})\). This implies the completeness of \(\{\psi _n\}_{n=0}^\infty \) in \({\mathscr {H}}_B({\mathbb {C}})\) stated in Theorem 4.2 since \(B_1\) is a Hilbert space isomorphism of \(L^2({\mathbb {R}})\) onto \({\mathscr {H}}_B({\mathbb {C}})\). For this reason, we have only to show (i) of Theorem 4.3. For this purpose, we need the following.
Lemma 4.4
Let \(\rho >0\) and let \(2\theta \in (-\pi /2,\pi /2)\). Then we have
Proof
The integrand is an even function of \(t\in {\mathbb {R}}\). By using change of variable \(t \mapsto \rho t\), we have
Let \(R>0\). Consider a contour \(\gamma _R\) which consists of \(\gamma _R=\gamma _R(1)\cup \gamma _R(2)\cup \gamma _R(3)\), where
Applying Cauchy’s theorem to the holomorphic function \(e^{-z^2}\) on \(\gamma _R\), we have
Here, we note that \(0<\cos (2\theta )\leqslant 1\) since \(2\theta \in (-\pi /2,\pi /2)\). Then, we deduce that
and
Then, we have
and
This completes the proof. \(\square \)
Finally, we complete the proof of Theorem 4.3.
Proof of Theorem 4.3
It suffices to show the part (i). We first compute the concrete form of \(\Psi _0(x)\). Recall the definition of \(\Psi _0(x)\)
where
Elementary computation gives
where
By using this and Lemma 4.4, we deduce that
which is desired.
Finally, we check the Rodrigues formula for \(\Psi _n\). By using the definition of \(\Psi _n\), we deduce that
This completes the proof. \(\square \)
References
Abreu, L.D., Dörfler, M.: An inverse problem for localization operators. Inverse Probl. 28, 115001 (2012)
Ali, S.T., Górska, K., Horzela, A., Szafraniec, F.H.: Squeezed states and Hermite polynomials in a complex variable. J. Math. Phys. 55, 012107 (2014)
Bargmann, V.: On a Hilbert space of analytic functions and an associated integral transform. Commun. Pure Appl. Math. 14, 187–214 (1961)
Chihara, H.: Bounded Berezin–Toeplitz operators on the Segal–Bargmann space. Integral Equ. Oper. Theory 63, 321–335 (2009)
Chihara, H.: Holomorphic Hermite functions and ellipses. Integral Transf. Spec. Funct. 28, 605–615 (2017)
Chihara, H.: Holomorphic Hermite functions in Segal–Bargmann spaces. Complex Anal. Oper. Theory 13, 351–374 (2019)
Daubechies, I.: Time-frequency localization operators: a geometric phase space approach. IEEE Trans. Inform. Theory 34, 605–612 (1988)
Folland, G.B.: Harmonic Analysis in Phase Space. Princeton University Press, Princeton (1989)
Parmeggiani, A.: Spectral theory of non-commutative harmonic oscillators: an introduction. In: Lecture Notes in Mathematics, vol. 1992. Springer, Berlin (2010)
Parmeggiani, A., Wakayama, M.: Non-commutative harmonic oscillators. I. Forum. Math. 14, 539–604 (2002)
Parmeggiani, A., Wakayama, M.: Non-commutative harmonic oscillators. II. Forum. Math. 14, 669–690 (2002)
Parmeggiani, A., Wakayama, M.: Corrigenda and remarks to: “non-commutative harmonic oscillators, I”. Forum. Math. 15, 955–963 (2003)
Sjöstrand, J.: Function spaces associated to global I-Lagrangian manifolds. In: Structure of Solutions to Differential Equations (Katata/Kyoto 1995), pp. 369–423, World Scientific Publishing, New York, (1996)
van Eijndhoven, S.J.L., Meyers, J.L.: New orthogonality relations for the Hermite polynomials and related Hilbert spaces. J. Math. Anal. Appl. 146, 89–98 (1990)
Yoshino, K.: Analytic continuation and applications of eigenvalues of Daubechies’ localization operator. Cubo 12, 203–212 (2010)
Author information
Authors and Affiliations
Corresponding author
Additional information
Communicated by Mohammad Sal Moslehian.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supported by the JSPS Grant-in-Aid for Scientific Research #19K03569.
Rights and permissions
About this article
Cite this article
Chihara, H. Bargmann-Type Transforms and Modified Harmonic Oscillators. Bull. Malays. Math. Sci. Soc. 43, 1719–1740 (2020). https://doi.org/10.1007/s40840-019-00771-3
Received:
Revised:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s40840-019-00771-3
Keywords
- Bargmann-type transforms
- Segal–Bargmann spaces
- Berezin–Toeplitz quantization
- Generalized Hermite functions
- Modified harmonic oscillators