Abstract
In this paper we investigated a relationship between the analytic generalized Fourier–Feynman transform associated with Gaussian process and the function space integral for exponential type functionals on the function space \(C_{a,b}[0,T]\). The function space \(C_{a,b}[0,T]\) can be induced by a generalized Brownian motion process. The Gaussian processes used in this paper are neither centered nor stationary.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
Let \(C_0[0,T]\) denote the classical Wiener space. In [4], Cameron and Storvick defined the “sequential” Feynman integral by means of finite dimensional approximations for functionals on the Wiener space \(C_0[0,T]\). The sequential definition for the Feynman path integral was intended to interpret the Feynman’s uniform measure [12] on continuous paths space \(C_0[0,T]\), because there is no countably additive measure as Lebesgue measure. It is well known that there is generally no quasi-invariant measure on infinite-dimensional linear spaces, see [13]. Thus, the Cameron and Storvick’s sequential Feynman integral is a rigorous mathematical formulation for the Feynman’s path integral. On the other hand, the concept of the “analytic” Feynman integral on the Wiener space \(C_0[0,T]\) was introduced by Cameron [1]. We refer to the reference [5, Section 1] for a heuristic structure of the analytic Feynman integral of functionals on \(C_0[0,T]\). The analytic Feynman integral is not defined in terms of a countably additive nonnegative measure. Rather, they are defined in terms of a process of analytic continuation and a limiting procedure. In this reason, Cameron and Storvick provided the Banach algebra \({\mathcal {S}}\) of analytic Feynman integrable functionals in [2]. Since then, in [3], Cameron and Storvick expressed the analytic Feynman integral of functionals in \({\mathcal {S}}\) as the limit of a sequence of Wiener integrals.
Let \(D=[0,T]\) and let \((\varOmega , {{\mathcal {B}}}, P)\) be a probability space. A generalized Brownian motion process (GBMP) on \(\varOmega \times D\) is a Gaussian process \(Y \equiv \{Y_t\}_{t\in D}\) such that \(Y_0=c\) almost surely for some constant \(c\in {\mathbb {R}}\) (in this paper we set \(c=0\)), and for any \(0\le s< t\le T\),
where \(N(m, \sigma ^2)\) denotes the normal distribution with mean m and variance \(\sigma ^2\), a(t) is a continuous real-valued function on [0, T], and b(t) is a monotonically increasing continuous real-valued function on [0, T]. Thus, the GBMP Y is determined by the functions a(t) and b(t). For more details, see [14, 15]. Note that when \(a(t)\equiv 0\) and \(b(t)=t\), the GBMP is a standard Brownian motion (Wiener process). We are obliged to point out that a standard Brownian motion is stationary in time, whereas a GBMP is generally not stationary in time, and is subject to a drift a(t).
In [8, 10], the authors defined the analytic generalized Feynman integral and the analytic generalized Fourier–Feynman transform (GFFT) on the function space \(C_{a,b}[0,T]\), and studied their properties and related topics. The function space \(C_{a,b}[0,T]\), induced by a GBMP, was introduced by Yeh in [14], and was used extensively in [5,6,7,8,9,10,11].
In this paper we extend the ideas of [3] to the functionals on the very general function space \(C_{a,b}[0,T]\). But our purpose of this paper is to obtain an expression of the analytic GFFT as a limit of a sequence of function space integrals on \(C_{a,b}[0,T]\). The result in this paper enables us that the analytic GFFTs of functionals on the function space \(C_{a,b}[0,T]\) can be interpreted as a limit of (non-analytic) function space transform.
The Wiener process used in [1,2,3,4] is centered and stationary in time and is free of drift. However, the Gaussian processes used in this paper, as well as in [6, 7], are neither centered nor stationary.
2 Preliminaries
In this section we first provide a brief background and some well-known results about the function space \(C_{a,b}[0,T]\) induced by the GBMP.
Let a(t) be an absolutely continuous real-valued function on [0, T] with \(a(0)=0\) and \(a'(t)\in L^2[0,T]\), and let b(t) be an increasing and continuously differentiable real-valued function with \(b(0)=0 \) and \(b'(t) >0\) for each \(t \in [0,T]\). The GBMP Y determined by a(t) and b(t) is a Gaussian process with mean function a(t) and covariance function \(r(s,t)=\min \{b(s),b(t)\}\). For more details, see [5, 7, 8, 10, 14, 15]. By Theorem 14.2 in [15], the probability measure \(\mu \) induced by Y, taking a separable version, is supported by \(C_{a,b}[0,T]\) (which is equivalent to the Banach space of continuous functions x on [0, T] with \(x(0)=0\) under the sup norm). Hence, \((C_{a,b}[0,T],{\mathcal {B}}(C_{a,b}[0,T]),\mu )\) is the function space induced by Y where \({\mathcal {B}}(C_{a,b}[0,T])\) is the Borel \(\sigma \)-field of \(C_{a,b}[0,T]\). We then complete this function space to obtain the measure space \((C_{a,b}[0,T],{\mathcal {W}}(C_{a,b}[0,T]),\mu )\) where \({\mathcal {W}}(C_{a,b}[0,T])\) is the set of all \(\mu \)-Carathéodory measurable subsets of \(C_{a,b}[0,T]\).
A subset B of \(C_{a,b}[0,T]\) is said to be scale-invariant measurable provided \(\rho B\) is \({\mathcal {W}}(C_{a,b}[0,T])\)-measurable for all \(\rho >0\), and a scale-invariant measurable set N is said to be scale-invariant null provided \(\mu (\rho N)=0\) for all \(\rho >0\). A property that holds except on a scale-invariant null set is said to hold scale-invariant almost everywhere (s-a.e.). A functional F is said to be scale-invariant measurable provided F is defined on a scale-invariant measurable set and \(F(\rho \,\,\cdot \,)\) is \({\mathcal {W}}(C_{a,b}[0,T])\)-measurable for every \(\rho >0\). If two functionals F and G defined on \(C_{a,b}[0,T]\) are equal s-a.e., we write \(F\approx G\).
Remark 2.1
The function space \(C_{a,b}[0,T]\) reduces to the Wiener space \(C_0[0,T]\), considered in papers [1,2,3,4] if and only if \(a(t)\equiv 0\) and \(b(t) = t\) for all \(t\in [0,T]\).
Let \(L_{a,b}^2[0,T]\) (see [8] and [10]) be the space of functions on [0, T] which are Lebesgue measurable and square integrable with respect to the Lebesgue–Stieltjes measures on [0, T] induced by \(a(\cdot )\) and \(b(\cdot )\); i.e.,
where \(|a|(\cdot )\) denotes the total variation function of \(a(\cdot )\). Then \(L_{a,b}^2[0,T]\) is a separable Hilbert space with inner product defined by
where \(m_{|a|,b}\) denotes the Lebesgue–Stieltjes measure induced by \(|a|(\cdot )\) and \(b(\cdot )\). In particular, note that \(\Vert u\Vert _{a,b}\equiv \sqrt{(u,u)_{a,b}} =0\) if and only if \(u(t)=0\) a.e. on [0, T]. For more details, see [8, 10].
Let
For \(w\in C_{a,b}'[0,T]\), with \(w(t)=\int _0^t z(s) d b(s)\) for \(t\in [0,T]\), let \(D: C_{a,b}'[0,T] \rightarrow L_{a,b}^2[0,T]\) be defined by the formula
Then \(C_{a,b}' \equiv C_{a,b}'[0,T]\) with inner product
is a separable Hilbert space. For more details, see [6, 7].
Note that the two separable Hilbert spaces \(L_{a,b}^2[0,T]\) and \(C_{a,b}'[0,T]\) are (topologically) homeomorphic under the linear operator given by Eq. (2.1). The inverse operator of D is given by
In the case that \(a(t)\equiv 0\), then the operator \(D:C_{0,b}'[0,T]\rightarrow L_{0,b}^2[0,T]\) is an isometry.
In this paper, in addition to the conditions put on a(t) above, we now add the condition
from which it follows that
where \(M=\sup _{t\in [0,T]}(1/b'(t))\). Thus, the function \(a: [0,T]\rightarrow {\mathbb {R}}\) satisfies the condition (2.2) if and only if \(a(\cdot )\) is an element of \(C_{a,b}'[0,T]\). Under the condition (2.2), we observe that for each \(w\in C_{a,b}'[0,T]\) with \(Dw=z\),
Let \(\{e_n\}_{n=1}^{\infty }\) be a complete orthonormal set in \((C_{a,b}'[0,T], \Vert \cdot \Vert _{C_{a,b}'})\) such that the \(De_n\)’s are of bounded variation on [0, T]. For \(w\in C_{a,b}'[0,T]\) and \(x\in C_{a,b}[0,T]\), we define the Paley–Wiener–Zygmund (PWZ) stochastic integral \((w,x)^{\sim }\) as follows:
if the limit exists.
We will emphasize the following fundamental facts. For each \(w\in C_{a,b}'[0,T]\), the PWZ stochastic integral \((w,x)^{\sim }\) exists for a.e. \(x\in C_{a,b}[0,T]\). If \(Dw=z\in L_{a,b}^2[0,T]\) is of bounded variation on [0, T], then the PWZ stochastic integral \((w,x)^{\sim }\) equals the Riemann–Stieltjes integral \(\int _0^T z(t)dx(t)\). Furthermore, for each \(w\in C_{a,b}'[0,T]\), \((w,x)^{\sim }\) is a Gaussian random variable with mean \((w,a)_{C_{a,b}'}\) and variance \(\Vert w\Vert _{C_{a,b}'}^2\). Thus, for an orthogonal set \(\{g_1,\ldots ,g_n\}\) of nonzero functions in \((C_{a,b}'[0,T],\Vert \cdot \Vert _{C_{a,b}'})\) and a Lebesgue measurable function \(f:{\mathbb {R}}^n \rightarrow {\mathbb {C}}\), it follows that
in the sense that if either side of Eq. (2.3) exists, both sides exist and equality holds. Also we note that for \(w, x\in C_{a,b}'[0,T]\), \((w,x)^{\sim }=(w,x)_{C_{a,b}'}\).
The following integration formula on the function space \(C_{a,b}[0,T]\) is also used in this paper:
for complex numbers \(\alpha \) and \(\beta \) with \(\mathrm{Re}(\alpha )> 0\).
3 Gaussian Processes
Let \(C_{a,b}^*[0,T]\) be the set of functions k in \(C_{a,b}'[0,T]\) such that Dk is continuous except for a finite number of finite jump discontinuities and is of bounded variation on [0, T]. For any \(w\in C_{a,b}'[0,T]\) and \(k\in C_{a,b}^*[0,T]\), let the operation \(\odot \) between \(C_{a,b}'[0,T]\) and \(C_{a,b}^*[0,T]\) be defined by
where DwDk denotes the pointwise multiplication of the functions Dw and Dk. Then \((C_{a,b}^*[0,T],\odot )\) is a commutative algebra with the identity b. For a more detailed study of the operation \(\odot \), see [7].
For each \(t\in [0,T]\), let \(\varPhi _t(\tau )=D^{-1}\chi _{[0,t]}(\tau )=\int _0^{\tau }\chi _{[0,t]}(u)db(u)\), \(\tau \in [0,T]\), and for \(k\in C_{a,b}'[0,T]\) with \(Dk\ne 0\) \(m_L\)-a.e. on [0, T] (\(m_L\) denotes the Lebesgue measure on [0, T]), let \({\mathcal {Z}}_{k}(x,t)\) be the PWZ stochastic integral
let \(\beta _k(t)=\int _0^t \{Dk (u)\}^2db(u)\), and let \(\alpha _k(t)=\int _0^t Dk (u)da(u)\). Then \({\mathcal {Z}}_k:C_{a,b}[0,T]\times [0,T]\rightarrow {\mathbb {R}}\) is a Gaussian process with mean function
and covariance function
In addition, by [15, Theorem 21.1], \({\mathcal {Z}}_{k}(\cdot ,t)\) is stochastically continuous in t on [0, T]. If Dk is of bounded variation on [0, T], then, for all \(x\in C_{a,b}[0,T]\), \({\mathcal {Z}}_{k}(x,t)\) is continuous in t. Also, for any functions \(k_1\) and \(k_2\) in \(C_{a,b}'[0,T]\),
Of course if \(k(t)\equiv b(t)\), then \({\mathcal {Z}}_b(x,t)=x(t)\), the continuous sample paths of the GBMP Y, of which the function space \(C_{a,b}[0,T]\) consists. Choosing \(a(t)\equiv 0\) and \(b(t)=t\) on [0, T], as commented in Remark 2.1 above, the function space \(C_{a,b}[0,T]\) reduces to the classical Wiener space \(C_0[0,T]\), and thus the Gaussian process (3.1) with \(k(t)\equiv t\) is an ordinary Wiener process.
From the properties of the PWZ stochastic integral and the operation \(\odot \) between \(C_{a,b}'[0,T]\) and \(C_{a,b}^*[0,T]\), it follows that for all \(\rho \in {\mathbb {R}}\),
and for any \(w\in C_{a,b}'[0,T]\) and each \(k\in C_{a,b}^*[0,T]\),
for \(\mu \)-a.e. \(x\in C_{a,b}[0,T]\). Thus, throughout the remainder of this paper, we require k to be in \(C_{a,b}^*[0,T]\) for each process \({\mathcal {Z}}_k\).
We define a class of those functions as follows: let
Then for any \(k\in \mathrm {Supp}_{C_{a,b}^*}[0,T]\), the Lebesgue–Stieltjes integrals
and
exist for all \(w\in C_{a,b}'[0,T]\).
4 Transforms on the Class of Exponential-Type Functionals
Given a function \(k\in \mathrm {Supp}_{C_{a,b}^*}[0,T]\), we define the generalized \({\mathcal {Z}}_k\)-function space integral (namely, the function space integral associated with the Gaussian paths \({\mathcal {Z}}_k(x,\cdot )\)) for functionals F on \(C_{a,b}[0,T]\) by the formula
Throughout this paper, let \({\mathbb {C}}\), \({\mathbb {C}}_+\) and \(\mathbb {{\widetilde{C}}}_+\) denote the set of complex numbers, complex numbers with positive real part, and non-zero complex numbers with nonnegative real part, respectively. Furthermore, for each \(\lambda \in {\mathbb {C}}\), \(\lambda ^{1/2}\) denotes the principal square root of \(\lambda \), i.e., \(\lambda ^{1/2}\) is always chosen to have nonnegative real part.
Definition 4.1
Given a function \(k\in \mathrm {Supp}_{C_{a,b}^*}[0,T]\), let \({\mathcal {Z}}_k\) be the Gaussian process given by (3.1) and let F be a \({\mathbb {C}}\)-valued scale-invariant measurable functional on \(C_{a,b}[0,T]\) such that
exists and is finite for all \(\lambda >0\). If there exists a function \(J_F^*({\mathcal {Z}}_k;\lambda )\) analytic on \({\mathbb {C}}_+\) such that \(J_F^*({\mathcal {Z}}_k;\lambda )=J_F({\mathcal {Z}}_k;\lambda )\) for all \(\lambda \in (0,+\infty )\), then \(J_F^* ({\mathcal {Z}}_k;\lambda )\) is defined to be the analytic \({\mathcal {Z}}_k\)-function space integral (namely, the analytic function space integral associated with the Gaussian paths \({\mathcal {Z}}_k(x,\cdot )\)) of F over \(C_{a,b}[0,T]\) with parameter \(\lambda \), and for \(\lambda \in {\mathbb {C}}_+\) we write
Let q be a non-zero real number and let F be a measurable functional whose analytic \({\mathcal {Z}}_k\)-function space integral \(J_F^*({\mathcal {Z}}_k;\lambda )\) exists for all \(\lambda \) in \({\mathbb {C}}_+\). If the following limit exists, we call it the analytic generalized \({\mathcal {Z}}_k\)-Feynman integral (namely, the analytic generalized Feynman integral associated with the paths \({\mathcal {Z}}_k(x,\cdot )\)) of F with parameter q and we write
where \(\lambda \) approaches \(-iq\) through values in \({\mathbb {C}}_+\).
Next we state the definition of the analytic GFFT associated with Gaussian process on function space.
Definition 4.2
Given a function \(k\in \mathrm {Supp}_{C_{a,b}^*}[0,T]\), let \({\mathcal {Z}}_k\) be the Gaussian process given by (3.1) and let F be a scale-invariant measurable functional on \(C_{a,b}[0,T]\) such that for all \(\lambda \in {\mathbb {C}}_+\) and \(y \in C_{a,b}[0,T]\), the following analytic \({\mathcal {Z}}_k\)-function space integral
exists. Let q be a non-zero real number. For \(p\in (1,2]\), we define the \(L_p\) analytic \({\mathcal {Z}}_k\)-GFFT (namely, the GFFT associated with the paths \({\mathcal {Z}}_k(x,\cdot )\)), \(T^{(p)}_{q,k}(F)\) of F, by the formula,
if it exists; i.e., for each \(\rho >0\),
where \(1/p +1/p' =1\). We define the \(L_1\) analytic \({\mathcal {Z}}_k\)-GFFT, \(T_{q,k}^{(1)}(F)\) of F, by the formula
for s-a.e. \(y\in C_{a,b}[0,T]\), if it exists.
We note that for \(1\le p \le 2\), \(T_{q,k}^{(p)}(F)\) is defined only s-a.e.. We also note that if \(T_{q,k}^{(p)}(F)\) exists and if \(F\approx G\), then \(T_{q,k}^{(p)}(G)\) exists and \(T_{q,k}^{(p)}(G)\approx T_{q,k}^{(p)}(F)\). Moreover, from Eqs. (4.2), (4.1), and (4.3), it follows that
in the sense that if either side exists, then both sides exist and equality holds.
Remark 4.3
Note that if \(k\equiv b\) on [0, T], then the analytic generalized \({\mathcal {Z}}_b\)-Feynman integral, \(I_{b}^{\mathrm { anf}_q}[F]\), and the \(L_p\) analytic \({\mathcal {Z}}_b\)-GFFT, \(T_{q, b}^{(p)}(F)\) agree with the previous definitions of the analytic generalized Feynman integral and the analytic GFFT respectively [5, 8, 10].
Let \({\mathcal {E}}\) be the class of all functionals which have the form
for some \(w\in C_{a,b}'[0,T]\) and for s-a.e. \(x\in C_{a,b}[0,T]\). More precisely, since we shall identify functionals which coincide s-a.e. on \(C_{a,b}[0,T]\), the class \({\mathcal {E}}\) can be regarded as the space of all s-equivalence classes of functionals of the form (4.5).
Given \(q\in {\mathbb {R}}\setminus \{0\}\), \(\tau \in C_{a,b}'[0,T]\), and \(k\in C_{a,b}^*[0,T]\), let \({\mathcal {E}}_{q,\tau ,k}\) be the class of all functionals having the form
for s-a.e. \(x\in C_{a,b}[0,T]\), where \(\varPsi _w\) is given by Eq. (4.5) and \(K^a_{q,\tau , k}\) is a complex number given by
The functionals given by Eq. (4.6) and linear combinations (with complex coefficients) of the \(\varPsi _{w}^{q,\tau ,k}\)’s are called the (partially) exponential-type functionals on \(C_{a,b}[0,T]\).
For notational convenience, let \(\varPsi _{w}^{0,\tau ,k} (x)=\varPsi _{w} (x)\) and let \({\mathcal {E}}_{0,\tau ,k}={\mathcal {E}}\). Then for any \((q,\tau ,k)\in {\mathbb {R}}\times C_{a,b}'[0,T]\times C_{a,b}^*[0,T]\), the class \({\mathcal {E}}_{q,\tau ,k}\) is dense in \(L_2(C_{a,b}[0,T])\), see [9, 11]. We then define the class \({\mathcal {E}}(C_{a,b}[0,T])\) to be the linear span of \({\mathcal {E}}\), i.e., \({\mathcal {E}}(C_{a,b}[0,T])=\text {Span}{\mathcal {E}}\).
Remark 4.4
-
(i)
One can see that \({\mathcal {E}}(C_{a,b}[0,T])=\text {Span}{\mathcal {E}}_{q,\tau ,k}\) for every \((q,\tau ,k)\in {\mathbb {R}}\times C_{a,b}'[0,T]\times C_{a,b}^*[0,T]\).
-
(ii)
The linear space \({\mathcal {E}} (C_{a,b}[0,T])\) is a commutative (complex) algebra under the pointwise multiplication and with identity \(\varPsi _0 \equiv 1\) because
$$\begin{aligned} \varPsi _{w_1}^{q_1,\tau _1,k_1} (x) \varPsi _{w_2}^{q_2,\tau _2,k_2} (x) =K^a_{q_1,\tau _1, k_1}K^a_{q_2,\tau _2, k_2}\varPsi _{w_1+w_2} (x) \end{aligned}$$for \(\mu \)-a.e. \(x\in C_{a,b}[0,T]\).
-
(iii)
Note that every exponential-type functional is scale-invariant measurable. Since we shall identify functionals which coincide s-a.e. on \(C_{a,b}[0,T]\), \({\mathcal {E}} (C_{a,b}[0,T])\) can be regarded as the space of all s-equivalence classes of exponential-type functionals.
The following two theorems are due to by Chang and Choi [6].
Theorem 4.5
Let \(\varPsi _w\in {\mathcal {E}}\) be given by Eq. (4.5). Then for all \(p\in [1,2]\), any non-zero real number q, and each function k in \(\mathrm {Supp}_{C_{a,b}^*}[0,T]\), the \(L_p\) analytic \({\mathcal {Z}}_{k}\)-GFFT of \(\varPsi _w\), \(T^{(p)}_{q,k}(\varPsi _w)\) exists and is given by the formula
where \(\varPsi _{w}^{q,w,k} \) is given by Eq. (4.6) with \(\tau \) replaced with w. Thus, \(T^{(p)}_{q,k}\!(\varPsi _w\!)\) is an element of \({\mathcal {E}}(\!C_{a,b}[0,T]\!)\).
Let F be a functional in \({\mathcal {E}}(C_{a,b}[0,T])\). Since \({\mathcal {E}}(C_{a,b}[0,T])=\text {Span}{\mathcal {E}}\), there exist a finite sequence \(\{w_1,\ldots , w_n\}\) of functions in \(C_{a,b}'[0,T]\), and a sequence \(\{c_1,\ldots , c_n\}\) in \({\mathbb {C}}\setminus \{0\}\) such that
Then for all \(p\in [1,2]\), any non-zero real number q, and each function k in \(\mathrm {Supp}_{C_{a,b}^*}[0,T]\), the \(L_p\) analytic \({\mathcal {Z}}_{k}\)-GFFT of F, \(T^{(p)}_{q,k}(F)\) exists and is given by the formula
where \(\varPsi _{w_j}^{q,w_j,k}\) is given by Eq. (4.6) with \(\tau \) and w replaced with \(w_j\) and \(w_j\), for each \(j\in \{1,\ldots ,n\}\), respectively.
Theorem 4.6
For all \(p\in [1,2]\), any \(q\in {\mathbb {R}}\setminus \{0\}\), and each \(k\in \mathrm {Supp}_{C_{a,b}^*}[0,T]\), the \(L_p\) analytic \({\mathcal {Z}}_{k}\)-GFFT, \(T_{q,k}^{(p)}: {\mathcal {E}}(C_{a,b}[0,T]) \rightarrow {\mathcal {E}}(C_{a,b}[0,T])\) is an onto transform.
5 Relationship Between the \({\mathcal {Z}}_k\)-Fourier–Feynman Transform and the Function Space Integral
In this section, we establish a relationship between the analytic \({\mathcal {Z}}_k\)-GFFT and the \({\mathcal {Z}}_k\)-function space integral of functionals in the class \({\mathcal {E}}(C_{a,b}[0,T])\).
Throughout this section, for convenience, we use the following notation: for \(\zeta \in \widetilde{{\mathbb {C}}}_+\) and \(n =1,2,\ldots \), let
where \(\{e_n\}_{n=1}^{\infty }\) is a complete orthonormal set of functions in \(C_{a,b}'[0,T]\).
Lemma 5.1
Let k be a function in \(\mathrm {Supp}_{C_{a,b}^*}[0,T]\), let \(\{e_{1},\ldots ,e_{n} \}\) be an orthonormal set of functions in \(C_{a,b}'[0,T]\), and let w be a function in \(C_{a,b}'[0,T]\). Then for each \(\zeta \in {\mathbb {C}}_+\), and \(n\in {\mathbb {N}}\), the functional \(\exp \big \{(w\odot k,x)^{\sim }\big \}G_n(\zeta ,x)\) is \(\mu \)-integrable, where \(G_n\) is given by (5.1). Also, it follows that
where
Proof
We note that given two functions \(k\in \mathrm {Supp}_{C_{a,b}^*}[0,T]\) and \(w\in C_{a,b}'[0,T]\), \(w\odot k\) is an element of \(C_{a,b}'[0,T]\). Using the Gram–Schmidt process, we obtain \(e_{n+1}^{w\odot k}\in C_{a,b}'[0,T]\) such that \(\{e_1,\ldots ,e_n, e_{n+1}^{w\odot k} \}\) forms an orthonormal set in \(C_{a,b}'[0,T]\) and
where
Next, for \(\zeta \in {\mathbb {C}}_+\), using (5.1), (2.3), the Fubini theorem, and (2.4), it follows that
as desired. \(\square \)
In our next theorem we express the analytic \({\mathcal {Z}}_k\)-GFFT of functionals in \({\mathcal {E}}(C_{a,b}[0,T])\) as the limit of a sequence of \({\mathcal {Z}}_k\)-function space integrals.
Theorem 5.2
Let \(F\in {\mathcal {E}} (C_{a,b}[0,T])\) be given by Eq. (4.9). Given a non-zero real number q, let \(\{\zeta _n \}\) be a sequence in \({\mathbb {C}}_+\) such that \(\zeta _n\rightarrow -iq\). Then, for all \(p\in [1,2]\), and each function k in \(\mathrm {Supp}_{C_{a,b}^*}[0,T]\), it follows that
for s-a.e. \(y\in C_{a,b}[0,T]\), where \(G_n\) is given by Eq. (5.1).
Proof
In view of Theorems 4.5 and 4.6 , it will suffice to show that Eq. (5.3) with p and F replaced with 1 and \(\varPsi _w\) holds true.
From Theorem 4.5, we know that the \(L_1\) analytic \({\mathcal {Z}}_k\)-GFFT of \(\varPsi _w\) given by (4.5), \(T_{q,k}^{(1)}(\varPsi _w)\), exists. Using (4.5), (3.2), the Fubini theorem, and the first expression of (5.2) with \(\zeta \) replaced with \(\zeta _n\), it follows that for all \(n\in {\mathbb {N}}\),
Next, using (5.4), (5.2), Parseval’s relation, (4.7), (4.5), (4.6) with \(\tau \) replaced with w, and (4.8) with \(p=1\), it follows that
for s-a.e. \(y\in C_{a,b}[0,T]\), as desired. \(\square \)
The following corollary follows immediately from (4.4) and (5.3).
Corollary 5.3
Let F, q and \(\{\zeta _n \}\) be as in Theorem 5.2. Then, for each function k in \(\mathrm {Supp}_{C_{a,b}^*}[0,T]\), it follows that
where \(G_n\) is given by Eq. (5.1).
We establish our next corollary after a careful examination of the proof of Theorem 5.2, and by using Eq. (4.1) instead of (4.4).
Corollary 5.4
Let \(F\in {\mathcal {E}} (C_{a,b}[0,T])\) be given by Eq. (4.9). Let \(\lambda \in {\mathbb {C}}_+\), and let \(\{\zeta _n \}\) be a sequence in \({\mathbb {C}}_+\) such that \(\zeta _n\rightarrow \lambda \). Then, for each function k in \(\mathrm {Supp}_{C_{a,b}^*}[0,T]\), it follows that
where \(G_n\) is given by Eq. (5.1).
Our next result, namely a change of scale formula for function space integrals, now follows easily by Corollary 5.4 above.
Corollary 5.5
Let \(F\in {\mathcal {E}} (C_{a,b}[0,T])\) be given by Eq. (4.9). Then for any \(\rho > 0\), and each function k in \(\mathrm {Supp}_{C_{a,b}^*}[0,T]\), it follows that
where \(G_n\) is given by Eq. (5.1).
Proof
Note that for every \(F\in {\mathcal {E}}(C_{a,b}[0,T])\) and all \(\rho >0\), the function space integral in the left-hand side of (5.6) exists. To ensure the equality in (5.6), simply choose \(\lambda = \rho ^{-2}\) and \(\zeta _n =\rho ^{-2}\) for every \(n\in {\mathbb {N}}\) in (5.5). \(\square \)
6 Concluding Remark
It is known that the class \({\mathcal {E}}(C_{a,b}[0,T])\) is a dense subspace of the space \(L_2(C_{a,b}[0,T])\). For a related work, see [9, 11]. Thus, using the \(L_2\)-approximation [11, Remark 4], one can develop the sequential approximation such as Eq. (5.3) for functionals F in \(L_2(C_{a,b}[0,T])\) whose \(L_p\) analytic \({\mathcal {Z}}_k\)-GFFT \(T_{q,k}^{(p)}(F)\) exists. But, there exists a (bounded) functional F in \(L_2(C_{a,b}[0,T])\) whose \(L_p\) \({\mathcal {Z}}_k\)-GFFT \(T_{q,k}^{(p)}(F)\) does not exist, see [5].
Indeed, the class \({\mathcal {E}}(C_{a,b}[0,T])\) is a very rich class of functionals on \(C_{a,b}[0,T]\). It contains many meaningful functionals which discussed in quantum mechanics. We finish this paper with a very simple example for such functionals, which arises in quantum mechanics.
Example 6.1
Consider the functional \(F_S\) given by
for s-a.e. \(x\in C_{a,b}[0,T]\). Then \(F_S\) is an element of \({\mathcal {E}}(C_{a,b}[0,T])\) because
where \(S: C_{a,b}'[0,T]\rightarrow C_{a,b}'[0,T]\) is an operator defined by
One can see that the adjoint operator \(S^*\) of S is given by
References
Cameron, R.H.: The ILSTOW and Feynman integrals. J. D’Anal. Math. 10, 287–361 (1962–63)
Cameron, R.H., Storvick, D.A.: Some Banach algebras of analytic Feynman integrable functionals. In: Analytic Functions (Kozubnik, 1979), Lecture Notes in Mathematics 798, pp. 18–67. Springer, Berlin (1980)
Cameron, R.H., Storvick, D.A.: Relationships between the Wiener integral and the analytic Feynman integral. Rend. Circ. Mat. Palermo 2(Suppl. 17), 117–133 (1987)
Cameron, R.H., Storvick, D.A.: A Simple Definition of the Feynman Integral With Applications. Mem. Amer. Math. Soc. 288, Amer. Math. Soc. (1983)
Chang, S.J., Choi, J.G.: Effect of drift of the generalized Brownian motion process: an example for the analytic Feynman integral. Arch. Math. 106, 591–600 (2016)
Chang, S.J., Choi, J.G.: Generalized transforms and generalized convolution products associated with Gaussian paths on function space. Commun. Pure Appl. Anal. 19, 371–389 (2020)
Chang, S.J., Choi, J.G., Ko, A.Y.: Multiple generalized analytic Fourier–Feynman transform via rotation of Gaussian paths on function space. Banach J. Math. Anal. 9, 58–80 (2015)
Chang, S.J., Choi, J.G., Skoug, D.: Integration by parts formulas involving generalized Fourier–Feynman transforms on function space. Trans. Am. Math. Soc. 355, 2925–2948 (2003)
Chang, S.J., Chung, H.S., Skoug, D.: Integral transforms of functionals in \(L^2(C_{a, b}[0, T])\). J. Fourier Anal. Appl. 15, 441–462 (2009)
Chang, S.J., Skoug, D.: Generalized Fourier-Feynman transforms and a first variation on function space. Integral Transforms Spec. Funct. 14, 375–393 (2003)
Chang, S.J., Skoug, D., Chung, H.S.: Relationships for modified generalized integral transforms, modified convolution products and first variations on function space. Integral Transforms Spec. Funct. 25, 790–804 (2014)
Feynman, R.P.: Space-time approach to non-relativistic quantum mechanics. Rev. Mod. Phys. 20, 367–387 (1948)
Yamasaki, Y.: Measures on infinite dimensional spaces. World Sci. Ser. Pure Math. 5, World Scientific Publishing, Singapore (1985)
Yeh, J.: Singularity of Gaussian measures on function spaces induced by Brownian motion processes with non-stationary increments. Ill. J. Math. 15, 37–46 (1971)
Yeh, J.: Stochastic Processes and the Wiener Integral. Marcel Dekker Inc., New York (1973)
Acknowledgements
The author would like to express his gratitude to the editor and the referees for their valuable comments and suggestions which have improved the original paper.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Choi, J.G. Relationship Between the Analytic Generalized Fourier–Feynman Transform and the Function Space Integral. Results Math 76, 108 (2021). https://doi.org/10.1007/s00025-021-01421-6
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s00025-021-01421-6
Keywords
- generalized Brownian motion process
- analytic generalized Fourier–Feynman transform
- Gaussian process
- exponential type functional