
==== Front
R Soc Open Sci
R Soc Open Sci
RSOS
royopensci
Royal Society Open Science
2054-5703
The Royal Society

rsos231579
10.1098/rsos.231579
100846659Mathematics
Research Articles
Uncertainty principles for coupled fractional Wigner–Ville distribution
Uncertainty principles for coupled fractional Wigner–Ville distribution
Nur Andi Tenri Ajeng 1 Conceptualization anditenriajeng09@gmail.com

https://orcid.org/0000-0003-1628-9163
Bahri Mawardi 1 Formal analysis mawardibahri@gmail.com

Bachtiar Nasrullah 1 Investigation nasrullahmipa.013@gmail.com

Rahim Amran 1 Validation amran@science.unhas.ac.id

1 Department of Mathematics, Hasanuddin University , Makassar 90245, Indonesia
5 2024
01 5 2024 May 1, 2024
01 5 2024 May 1, 2024
11 5 23157923 10 2023 October 23, 2023
02 2 2024 February 2, 2024
22 2 2024 February 22, 2024
© 2024 The Authors.
2024
https://creativecommons.org/licenses/by/4.0/ Published by the Royal Society under the terms of the Creative Commons Attribution License http://creativecommons.org/licenses/by/4.0/, which permits unrestricted use, provided the original author and source are credited.

The coupled fractional Wigner–Ville distribution is a more general version of the fractional Wigner–Ville distribution. Main properties including boundedness, Moyal’s formula and inversion formula are studied in detail for the transformation. Additionally, the relation of the coupled fractional Wigner–Ville distribution with the two-dimensional Fourier transform is studied. We also present the relationship between the coupled fractional Wigner–Ville distribution with the two-dimensional Wigner–Ville distribution. We show how the properties and relations allow us to derive several versions of the uncertainty inequalities related to the coupled fractional Wigner–Ville distribution.

coupled fractional Fourier transform
; uncertainty principle
; fractional Fourier transform
==== Body
pmc1. Introduction

In a series of articles [1–8], the fractional Fourier transform has become a standard mathematical tool in a large number of areas including quantum mechanics, neural networks, differential equations, optics, pattern recognition, radar, sonar, and other communication systems. It can be understood as an expansion of the Fourier transform, which was first introduced in 1980 by Namias [9]. In the latest work of the authors [10,11], the coupled fractional Fourier transform has been proposed. The generalized version can be considered as a variant of the two-dimensional fractional Fourier transform. Based on the kernel of the coupled fractional Fourier transform, the authors [12] have constructed the short-time coupled fractional Fourier transform. Later in [13], the authors proposed the coupled fractional Wigner–Ville distribution (CFrWVD), which is a natural generalization of the fractional Wigner–Ville distribution [14] and the classical Wigner–Ville distribution.

Though several essential properties of this generalized transformation have been investigated in detail [13], and the uncertainty principles were also reported [15], several uncertainty principles associated with this transformation such as sharp Hausdorff–Young inequality do not seem to have been realized so far. In contrast, the proof of the uncertainty principles is based on the definition and properties of the CFrWVD, and we implement the relation between the CFrWVD and two-dimensional Fourier transform, the proof of which is simpler. Furthermore, our main results may be viewed as a continuation of the results in [15].

In the present study, we deal with the CFrWVD. Our main contribution is to explore several versions of the uncertainty inequalities concerning the CFrWVD, which is one of the fundamental results related to the transformation. To arrive at the results, we introduce a definition of the CFrWVD and investigate the main properties. We also provide a direct connection between the CFrWVD and the two-dimensional Fourier transform.

The organization of the work is as follows. In §2, we collect some essential facts on the fractional Fourier transform and the coupled fractional Fourier transform. Section 3 concentrates on the derivation of the main properties of the CFrWVD. We also demonstrate its relation with the two-dimensional Fourier transform, which will be useful to obtain some inequalities related to the CFrWVD. Section 4 is devoted to the derivation of some uncertainty principles concerning the CFrWVD. Lastly, in §5, we conclude.

2. Preliminaries

First of all, we recall the basic facts related to the fractional Fourier transform (FrFT) and the coupled fractional Fourier transform and their basic properties, which will be needed in the sequel. We also introduce a definition of the CFrWVD. We begin by recalling the well-known definition below.

Definition 2.1. We define the space of measurable functions on ℝ2 , such that

(2.1) ‖f‖Lr(R2)=(∫  R2|f(τ)|rdτ)1/r<∞,1≤r<∞.

Here, τ=(τ1,τ2)∈R2,dτ=dτ1dτ2.

Especially, for r→∞ , we get

(2.2) ‖f‖L∞(R2)=esssupτ∈R2|f(τ)|.

Furthermore, if f is continuous, then equation (2.2) changes to

(2.3) ∥f∥L∞⁢(ℝ2)=sup𝝉∈ℝ2⁡|f⁢(𝝉)|.

The usual inner product of L2⁢(ℝ2) is then defined as

(2.4) ⟨f,g⟩L2(R2)=∫  R2f(τ)g(τ)¯dτ.

Let us now introduce a definition of the two-dimensional FrFT.

Definition 2.2. The two-dimensional FrFT with parameter θ is defined for a function f∈L1⁢(ℝ2) by [3,8]

(2.5) Fθ{f}(η)=∫  R2f(τ)Kθ(η,τ)dτ,

where the kernel function Kθ⁢(𝛈,𝛕) is given by

(2.6) Kθ(η,τ)={Aθei(|τ|2+|η|2)cot⁡θ2−iτ⋅η csc⁡θ, θ≠nπδ(τ−η), θ=2nπδ(τ+η), θ=(2n+1)π,n∈Z,

for which the Dirac delta function δ⁢(𝛕-𝛈)=δ⁢(τ1-η1)⁢δ⁢(τ2-η2) and

(2.7) Aθ=1-i⁢cot⁡θ2⁢π,Aθ¯=1+i⁢cot⁡θ2⁢π.

It is straightforward to verify that the FrFT kernel fulfils the following basic properties:

Kθ(η,τ)¯=K−θ(η,τ)

and

∫R2Kθ(η,τ)Kθ(η′,τ)¯dτ=δ(η−η′),

where Kθ⁢(𝜼,𝝉)¯ stands for the complex conjugate of Kθ⁢(𝜼,𝝉) .

Definition 2.3. Suppose that f∈L1⁢(ℝ2) and ℱθ⁢{f}∈L1⁢(ℝ2) . The inverse transform of the two-dimensional FrFT of the function f is given by the integral [3,8]

(2.8) f(τ)=Fθ−1[Fθ{f}](τ)=∫  R2Fθ{f}(η)Aθ¯e−i(|τ|2+|η|2)cot⁡θ2−iτ⋅η  csc⁡θdη.

Equation (2.8) shows how to recover the function from its FrFT. The following definition is important in this article.

Definition 2.4. (CFrFT definition). The coupled fractional Fourier transform (CFrFT) is defined for any function f∈L1⁢(ℝ2)∩L2⁢(ℝ2) by [10,11]

(2.9) Fα,β{f}(η)=∫  R2f(τ)Kα,β(η,τ)dτ=∫  R2f(τ)d(γ)e−i(a(γ)(|τ|2+|η|2)−τ⋅Mη)dτ.

In this case,

(2.10) Kα,β(η,τ)=d(Υ)e−i(α(Υ)(∣τ∣2+∣η∣2)−τ.Mη).

Here,

γ=α+β2, δ=α−β2, a=(γ)=cot⁡γ2, b=(γ,δ)=cos⁡δsin⁡γ,c(γ,δ)=sin⁡δsin⁡γ, d(γ)=ie−iγ2πsin⁡γ, M=(b(γ,δ)c(γ,δ)−c(γ,δ)b(γ,δ)),

with α,β∈R such that α+β∉2πZ.

Equation (2.9) may be expressed in the form

(2.11) Fα,β{f}(η)=d(γ)∫  R2(f(τ)e−ia(γ)|τ|2)e−ia(γ)|η|2eiτ⋅Mηdτ.

Denoting

(2.12) g⁢(𝝉)=f⁢(𝝉)⁢e-i⁢a⁢(γ)⁢|𝝉|2,

we obtain

|g⁢(𝝉)|=|f⁢(𝝉)|.

Due to equation (2.12), equation (2.11) takes the following form:

(2.13) (d(γ))−1eia(γ)|η|2Fα,β{f}(η)=∫  R2g(τ)eiτ⋅Mηdτ=F{g}(−Mη).

This equation describes the basic connection between the CFrFT and the two-dimensional Fourier transform. Here, ℱ⁢{f} stands for the two-dimensional Fourier transform of f∈L2⁢(ℝ2) given by [16–18]

(2.14) F{f}(η)=∫  R2f(τ)e−iτ⋅ηdτ.

The function f⁢(𝝉) in equation (2.14) can be obtained in terms of the CFrFT using the following definition.

Definition 2.5. For every f∈ℱα,β⁢{f},L1⁢(ℝ2) , the inverse of the CFrFT is given by [10,11]

(2.15) f(τ)=∫  R2Fα,β{f}(η)Kα,β(η,τ)¯dη=d(γ)¯∫  R2Fα,β{f}(η)ei(a(γ)(|τ|2+|η|2)−τ⋅Mη)dη.

A Parseval identity is valid for the CFrFT. For any f,g∈L2⁢(ℝ2) , the following relation is satisfied:

(2.16) ⟨f,g⟩L2(R2)=⟨Fα,β{f},Fα,β{g}⟩L2(R2)

and

(2.17) ∥f∥L2⁢(ℝ2)2=∥ℱα,β⁢{f}∥L2⁢(ℝ2)2.

Let us recall a definition of the CFrWVD [13]. It is constructed by replacing the kernel Fourier in the definition of the two-dimensional Wigner–Ville distribution with the kernel function of the CFrFT.

Definition 2.6. The CFrWVD is defined for functions f,g∈L2⁢(ℝ2) by [13]

(2.18) Wf,gα,β(x,η)=∫  R2f(x+τ2) g(x−τ2)¯d(γ)e−i(a(γ)(|τ|2+|η|2)−τ⋅Mη)dτ.

3. Coupled fractional Wigner–Ville distribution and main properties

In this section, we investigate the essential properties of the CFrWVD. The properties will be used in the later part of this article.

From equation (2.18), we get

(3.1) Wf,gα,β(x,η)=∫  R2hf,g(x,τ)Kα,β(η,τ)dτ=Fα,β{hf,g(x,τ)}(η),

where Kα,β⁢(𝜼,𝝉) is defined by equation (2.10) and

(3.2) hf,g(x,τ)=f(x+τ2)g(x−τ2)¯.

Due to equation (2.18), we have

Wf,gα,β(x,η)=∫  R2hf,g(x,τ)d(γ)e−i(a(γ)(|τ|2+|η|2)−τ⋅Mη)dτ.

The above equation may be expressed as

d−1(γ)eia(γ)|η|2Wf,gα,β(x,η)=∫  R2hf,g(x,τ)e−ia(γ)|τ|2eiτ⋅Mηdτ.

Applying equation (2.14) to the above identity gives

(3.3) d−1(γ)eia(γ)|η|2Wf,gα,β(x,η)=F{hˇf,g(x,τ)}(−Mη).

This equation is equal to

(3.4) Wf,gα,β(x,η)=F{hˇf,g(x,τ)}(−Mη)d(γ)e−ia(γ)|η|2,

where

(3.5) hˇf,g(x,τ)=f(x+τ2) g(x−τ2)¯e−ia(γ)|τ|2.

Equation (3.4) describes a direct interaction between the CFrWVD and the two-dimensional Fourier transform. It plays a crucial for deriving the main results in this article.

Further, from definition 2.6, we have

(3.6) Wf,gα,β(x,η)=∫  R2f(x+τ2) g(x−τ2)¯d(γ)e−i(a(γ)(|τ|2+|η|2)−τ⋅Mη)dτ=d(γ)e−ia(γ)|η|2∫  R2f(x+τ2) g(x−τ2)¯e−ia(γ)|τ|2e−τ⋅Mηdτ=d(γ)e−ia(γ)|η|2Wfˇ,g(x,Mη),

where fˇ⁢(𝒙+𝝉2)=f⁢(𝒙+𝝉2)⁢e-i⁢a⁢(γ)⁢|𝝉|2 and Wfˇ,g⁢(𝒙,M⁢𝜼) are the two-dimensional Wigner–Ville distribution [16,17]. Equation (3.6) above explains the direct relation of the CFrWVD to the two-dimensional Wigner–Ville distribution.

Some important properties of the CFrWVD above are as follows.

Theorem 3.1. (Boundedness). Let f,g∈L2⁢(ℝ2) , then we have

(3.7) |𝒲f,gα,β⁢(𝒙,𝜼)|2≤4π2⁢|sin⁡γ|2⁢∥f∥L2⁢(ℝ2)2⁢∥g∥L2⁢(ℝ2)2.

Proof. Thanks to the Cauchy–Schwarz inequality, we obtain

|Wf,gα,β(x,η)|2=|∫  R2f(x+τ2) g(x−τ2)¯d(γ)e−i(a(γ)(|τ|2+|η|2)−τ⋅Mη)dτ|2≤∫  R2|f(x+τ2) g(x−τ2)¯d(γ)|2dτ=|d(γ)|2∫  R2|f(x+τ2)|2dτ∫  R2|g(x−τ2)¯|2dτ=4π2|sin⁡γ|2‖f‖L2(R2)2‖g‖L2(R2)2.

This shows that 𝒲f,gα,β⁢(𝒙,𝜼) is bounded on L2⁢(ℝ2) .

Theorem 3.2. (Moyal’s formula). For any functions f1,f2∈L2⁢(ℝ2) and g1,g2∈L2⁢(ℝ2) . Then, one has

(3.8) ∫  R2∫  R2Wf1,g1α,β(x,η)Wf2,g2α,β(x,η)¯dηdx=4⟨f1,f2⟩L2(R2)⟨g2,g1⟩L2(R2).

Proof. With the aid of equations (2.16), (3.1) and (3.2), we find that

(3.9) ∫  R2∫  R2Wf1,g1α,β(x,η)Wf2,g2α,β(x,η)¯dηdx=∫  R2∫  R2Fα,β{hf1,g1(x,τ)}(η)Fα,β{hf2,g2(x,τ)}(η)¯dηdx=∫  R2∫  R2hf1,g1(x,τ)hf2,g2(x,τ)¯dτdx=∫  R2∫  R2f1(x+τ2) g1(x−τ2)¯f2(x+τ2) g2(x−τ2)¯¯dτdx=∫  R2∫  R2f1(x+τ2) g1(x−τ2)¯ g2(x−τ2)f2(x+τ2)¯dτdx.

We use Fubini’s theorem to obtain

∫  R2∫  R2Wf,gα,β{f1}(x,η)Wf,gα,β{f2}(x,η)¯dηdx=∫  R2f1(x+τ2)f2(x+τ2)¯dτ∫  R2g1(x−τ2)¯ g2(x−τ2)dx=4⟨f1,f2⟩L2(R2)⟨g2,g1⟩L2(R2),

which proves equation (3.8).

An immediate consequence of the above theorem is the following:

f g1=g2=g then

(3.10) ∫  R2∫  R2Wf1,gα,β(η,x)Wf2,gα,β(x,η)¯dηdx=4‖g‖L2(R2)2⟨f1,f2⟩L2(R2).

If f1=f2 and g1=g2 then

(3.11) ∫  R2∫  R2|Wf1,g1α,β(η,x)|2dηdx=4‖g1‖L2(R2)2‖f1‖L2(R2)2.

Theorem 3.3. (Inversion formula). Let f,g∈L2⁢(ℝ2) be two functions. Then, every f∈L2⁢(ℝ2) we have

(3.12) f(τ)=1g(0)∫  R2Wf,gα,β(τ2,η)Kα,β(η,τ)¯dη.

Proof. From equations (3.1) and (3.2), we get

(3.13) f(x+τ2) g(x−τ2)¯=∫  R2Wf,gα,β(x,η)Kα,β(η,τ)¯dη.

If we put 𝒙=𝝉2 , the above identity is turned into

f(τ)g(0)¯=∫  R2Wf,gα,β(τ2,η)Kα,β(η,τ)¯dη,

which completes the proof.

Remark 3.4. It should be noticed that theorems 3.2 and 3.3 in the present work differ in terms of constants from the ones proposed in [13].

To motivate the need for the CFrWVD mentioned earlier, we will look at the examples below.

Example 3.5. Find the CFrWVD of functions f and g , given by

(3.14) f⁢(𝝉)=e-|𝝉|2

and

(3.15) g⁢(𝝉)={1,for⁢  0≤τ1,τ2≤10,τ1,τ2 elsewhere.

Solution. It follows from equation (2.18) that

(3.16) Wf,gα,β(x,η)=d(γ)∫  R2e−|x+τ2|2g(x−τ2)e−i(a(γ)(|τ|2+|η|2)−τ⋅Mη)dτ=d(γ)∫  R2e−((x1+τ12)2+(x2+τ22)2)g(x−τ2)×e−i(a(γ)(τ12+τ22+η12+η22)−τ1(bη1+cη2)−τ2(bη2−cη1)dτ1dτ2=d(γ)∫2(x1−1)2x1∫2(x2−1)2x2e−(x1+τ12)2−ia(γ)τ12+τ1(bη1+cη2)=d(γ)e−ia(γ)(η12+η22)+(x12+x22)∫2(x1−1)2x1e−τ124−ia(γ)τ12+τ1(bη1+cη2)−τ1x1dτ1×∫2(x2−1)2x2e−τ224−ia(γ)τ22+τ2(bη2+cη1)−τ2x2dτ2.

Furthermore, we get

(3.17) 𝒲f,gα,β⁢(𝒙,𝜼)=d⁢(γ)⁢e-i⁢a⁢(γ)⁢(η12+η22)+(x12+x22)⁢∫2⁢(x1-1)2⁢x1e(14+i⁢a⁢(γ))⁢τ12+τ1⁢(b⁢η1+c⁢η2-x1)⁢𝑑τ1×∫2⁢(x2-1)2⁢x2e(14+i⁢a⁢(γ))⁢τ22+τ2⁢(b⁢η2+c⁢η1-x2)dτ2=d⁢(γ)⁢e-i⁢a⁢(γ)⁢(η12+η22)+(x12+x22)⁢∫2⁢(x1-1)2⁢x1e(14+i⁢a⁢(γ))⁢(τ12-τ1⁢b⁢η1+c⁢η2-x114+i⁢a⁢(γ))⁢𝑑τ1×∫2⁢(x2-1)2⁢x2e(14+i⁢a⁢(γ))⁢(τ22-τ2⁢b⁢η2+c⁢η1-x214+i⁢a⁢(γ))dτ2.

Hence,

(3.18) Wf,gα,β(x,η)=d(γ)e−ia(γ)(η12+η22)+(x12+x22)∫2(x1−1)2x1e(14+ia(γ))(τ1−(bη1+cη2−x1)2(14+ia(γ)))2e−(bη1+cη2−x1)24(14+ia(γ))dτ1×∫2(x2−1)2x2e(14+ia(γ))(τ2−(bη2+cη1−x2)2(14+ia(γ)))2e−(bη2+cη1−x2)24(14+ia(γ))dτ2=d(γ)e−ia(γ)(η12+η22)+(x12+x22)−(bη1+cη2−x1)21+i4a(γ)−(bη2+cη1−x2)21+i4a(γ)×∫2(x1−1)2x1e−(14+ia(γ)τ1−(bη1+cη2−x1)214+ia(γ))2dτ1∫2(x2−1)2x2e−(14+ia(γ)τ2−(bη2+cη1−x2)214+ia(γ))2dτ2.

If we set

(3.19) u=(bη1+cη2−x1)214+ia(γ)−14+ia(γ)τ1

and

(3.20) z=(b⁢η2+c⁢η1-x2)2⁢14+i⁢a⁢(γ)-14+i⁢a⁢(γ)⁢τ2,

then we get

(3.21) Wf,gα,β(x,η)=π1+i4a(γ)d(γ)e−ia(γ)(η12+η22)−(x12+x22)−(bη1+cη2−x1)21+i4a(γ)−(bη2+cη1−x2)21+i4a(γ)[(erf⁡((bη1+cη2−x1)214+ia(γ)+214+ia(γ)(x1−1))−erf⁡((bη1+cη2−x1)214+ia(γ)−2x114+ia(γ)))×(erf⁡((bη2+cη1−x2)214+ia(γ)+214+ia(γ)(x2−1))−erf⁡((bη2+cη1−x2)214+ia(γ)−2x214+ia(γ)))],

where

erf⁡z=2π⁢∫0ze-t2⁢𝑑t,for⁢all⁢z.

We plot example 3.5 in figures 1 and 2.

Figure 1. (a) Real part and (b) imaginary part of the coupled fractional Wigner–Ville distribution of example 3.5 in the spatial domain ( 𝒙 domain) for α=π4 , β=π2𝜼=5 and 𝒙=[-5,5] .

Real part and imaginary part of the coupled fractional Wigner–Ville distribution in the spatial domain.

Figure 2. (a) Real part and (b) imaginary part of the coupled fractional Wigner–Ville distribution of example 3.5 in the frequency domain ( 𝜼 domain) for α=π4 , β=π2𝒙=5 and 𝜼=[-5,5] .

Real part and imaginary part of the coupled fractional Wigner–Ville distribution in the frequency domain.

Example 3.6. Find the CFrWVD of the functions f and g , defined by

(3.22) f(τ)=g(τ)=e−|τ|2.

Solution. Substituting equation (3.22) into equation (2.18), we obtain

(3.23) Wf,gα,β(x,η)=d(γ)∫  R2e−|x+τ2|2e−|x−τ2|2e−i(a(γ)(|τ|2+|η|2)−τ1(bη1+cη2)−τ2(bη2−cη1))dτ1dτ2=d(γ)∫  R2e−((x1+τ12)2+(x2+τ22)2)e−((x1−τ12)2+(x2−τ22)2)×e−i(a(γ)(τ12+τ22+η12+η22)−τ1(bη1+cη2)−τ2(bη2−cη1)dτ1dτ2=d(γ)∫  R2e−((x1+τ12)2+(x1−τ12)2)e−i(a(γ)τ12−τ1(bη1+cη2))e−((x2+τ22)2+(x2−τ22)2)×e−i(a(γ)τ22−τ2(bη1+cη2))e−ia(γ)(η12+η22)dτ1dτ2=d(γ)e−ia(γ)(η12+η22)∫  Re−(2(x12+(τ12)2))−i(a(γ)τ12−τ1(bη1+cη2))dτ1×∫  Re−(2(x22+(τ22)2))−i(a(γ)τ22−τ1(bη2+cη1))dτ2.

Furthermore, we get

(3.24) Wf,gα,β(x,η)=d(γ)e−ia(γ)(η12+η22)−2(x12+x22)∫  R2e−τ122−i(a(γ)τ12+τ1(bη1+cη2))dτ1×∫  R2e−τ222−i(a(γ)τ22+τ2(bη2−cη1))dτ2=d(γ)e−ia(γ)(η12+η22)−2(x12+x22)∫  R2e−(12+ia(γ))τ12+τ1(bη1+cη2)dτ1×∫  R2e−(12+ia(γ))τ22+τ2(bη2−cη1)dτ2.

Equation (3.24) can be rewritten as

(3.25) Wf,gα,β(x,η)=d(γ)e−ia(γ)(η12+η22)−2(x12+x22)∫  Re−(12+ia(γ))(τ12+τ1(bη1+cη2)(12+ia(γ)))dτ1×∫  Re−(12+ia(γ))(τ22+τ2(bη2−cη1)(12+ia(γ)))dτ2=d(γ)e−ia(γ)(η12+η22)−2(x12+x22)∫  Re−(12+ia(γ))((τ1+(bη1+cη2)(1+i2a(γ)))2−((bη1+cη2)(1+i2a(γ)))2)dτ1×∫  Re−(12+ia(γ))((τ2+(bη2−cη1)(1+i2a(γ)))2−((bη2−cη1)(1+i2a(γ)))2)dτ2=d(γ)e−ia(γ)(η12+η22)−2(x12+x22)+(12+ia(γ))((bη1+cη2)(1+i2a(γ)))2+(12+ia(γ))((bη2−cη1)(1+i2a(γ)))2×∫  Re−(12+ia(γ))(τ1+bη1+cη21+i2a(γ))2dτ1∫  Re−(12+ia(γ))(τ2+bη2−cη11+i2a(γ))2dτ2=d(γ)e−ia(γ)(η12+η22)−2(x12+x22)+(bη1+cη2)22(1+i2a(γ))+(bη2−cη1)22(1+i2a(γ))∫  Re−(1+i2a(γ)2τ1+2(bη1+cη2)1+i2a(γ))2dτ1×∫  Re−(1+i2a(γ)2τ2+2(bη2−cη1)1+i2a(γ))2dτ2.

We finally arrive at

(3.26) 𝒲f,gα,β⁢(𝒙,𝜼)=2⁢π⁢d⁢(γ)1+i⁢2⁢a⁢(γ)⁢e-i⁢a⁢(γ)⁢(η12+η22)-2⁢(x12+x22)+(b⁢η1+c⁢η2)2+(b⁢η2-c⁢η1)22⁢(1+i⁢2⁢a⁢(γ)).

We plot example 3.6 in figures 3 and 4.

Figure 3. (a) Real part and (b) imaginary part of the coupled fractional Wigner–Ville distribution of example 3.6 in the frequency domain ( 𝜼 domain) for α=π4 , β=π2𝒙=5 and 𝜼=[-5,5] .

Real part and imaginary part of the coupled fractional Wigner–Ville distribution in the frequency domain.

Figure 4. (a) Real part and (b) imaginary part of the coupled fractional Wigner–Ville distribution of example 3.6 in the spatial domain ( 𝒙 domain) for α=π4 , β=π2𝜼=5 and 𝒙=[-5,5] .

Real part and imaginary part of the coupled fractional Wigner–Ville distribution in the spatial domain.

4. Uncertainty principles for coupled fractional Wigner–Ville distribution

An uncertainty principle offers information about a signal and its Fourier transform in the time–frequency plane. More precisely, it states that a signal and its Fourier transform cannot simultaneously concentrate around a point. It is known that the most important property of any generalized transformation is the uncertainty principle. Therefore, various uncertainty principles of different types of transformations have been proposed [19–25]. In this section, we explore several versions of the uncertainty principles in the context of the CFrWVD.

4.1. Heisenberg uncertainty principle

Here, we shall expand the idea of the Heisenberg uncertainty principle for the two-dimensional Fourier transform to that of the CFrWVD. In this respect, we shall state and prove the following theorem.

Theorem 4.1. Let two functions f,g∈L1⁢(ℝ2)∩L2⁢(ℝ2) . Then, one has

(4.1) ‖f‖L2(R2)≤16π2|sin⁡γ|4‖g‖L2(R2)∫  R2|τ−x|2|f(τ)|2dτ∫R2∫  R2|η|2|Wf,gα,β(x,η)|2dηdx.

Proof. By virtue of the uncertainty principle for the two-dimensional Fourier transform, we have

(4.2) ∫  R2|τ|2|f(τ)|2dτ∫  R2|η|2|F{f}(η)|2dη≥14(∫  R2|f(τ)|2dτ)12.

Furthermore, we obtain

(4.3) ∫  R2|τ|2|hˇf,g(x,τ)|2dτ∫  R2|η|2|F{hˇf,g(x,τ)}(η)|2dη≥14(∫  R2|hˇf,g(x,τ)|2dτ)12.

Substituting 𝜼 for -M⁢𝜼 in the above equation yields

(4.4) ∫  R2|τ|2|hˇf,g(x,τ)|2dτ∫  R2|(detM)|3|η|2|F{hˇf,g(x,τ)}(−Mη)|2dη≥14(∫  R2|hˇf,g(x,τ)|2dτ)12.

From equations (3.3) and (3.5) it will lead to

(4.5) |detM|3|d−1(γ)|2∫  R2|τ|2|f(x+τ2)g(x−τ2)¯|2dτ∫  R2|η|2|eia(γ)|η|2Wf,gα,β(x,η)|2dη≥14(∫  R2|f(x+τ2)g(x−τ2)¯|2dτ)12.

Integrating both sides of equation ( 4.5 ) with respect to d⁢𝒙 , we obtain

(4.6) 14∫  R2∫  R2|f(x+τ2)g(x−τ2)¯|2dτdx≤4π2|sin⁡γ|4∫  R2∫  R2|τ|2|f(x+τ2)g(x−τ2)¯|2dτdx∫  R2∫  R2|η|2|Wf,gα,β(x,η)|2dηdx.

Fubini’s theorem gives

(4.7) (∫  R2|f(τ)|2dτ)12(∫  R2|g(x)|¯2dx)12≤16π2|sin⁡γ|4∫  R2|2(τ−x)|2|f(τ)|2dτ∫  R2|g(x)|2dx∫  R2∫  R2|η|2|Wf,gα,β(x,η)|2dηdx.

Hence,

‖f‖L2(  R2)≤16π2|sin⁡γ|4‖g‖L2(  R2)∫  R2|τ−x|2|f(τ)|2dτ∫  R2∫  R2|η|2|Wf,gα,β(x,η)|2dηdx,

which proves the theorem.

4.2. Sharp Hausdorff–Young inequality

The purpose of this part is to build sharp Hausdorff–Young inequality related to the CFrWVD. This principle generalizes sharp Hausdorff–Young inequality for the two-dimensional Fourier transform to the CFrWVD. This principle is very useful in deriving Lieb’s inequality related to the proposed CFrWVD.

Theorem 4.2. Let p∈[1,2] , such that 1p+1q=1 , then for any f,g∈L2⁢(ℝ2) , there holds

(4.8) (∫  R2∫  R2|Wf,gα,β(x,η)pdηdx|)1p≤41qsin |γ|2p−12πC(p)‖f‖Lq(R2)‖g‖Lq(R2),

where

(4.9) C⁢(p)=p1p⁢q-1q.

Proof. By virtue of sharp Hausdorff–Young inequality for the two-dimensional Fourier transform, it follows that

(4.10) (∫  R2|F{f}(η)|pdη)1p≤C(p)(∫  R2|f(τ)|qdτ)1q.

Inserting f⁢(𝝉) by hˇf,g⁢(𝒙,𝝉) to both sides of equation (4.10) results in

(4.11) (∫  R2|F{hˇf,g(x,τ)(η)}|pdη)1p≤C(p)(∫  R2|hˇf,g(x,τ)|qdτ)1q.

Substituting 𝜼 for -M⁢𝜼 in equation (4.11), it is turned into

(4.12) (∫  R2|F{hˇf,g(x,τ)(−Mη)}|pd(−Mη))1p≤C(p)(∫  R2|f(x+τ2)g(x−τ2)¯e−ia(γ)|τ|2|qdτ)1q.

Due to equation (3.3), we obtain

(4.13) 2π|sin⁡γ||sin⁡γ|2p(∫  R2|Wf,gα,β(x,η)|pdη)1p≤C(p)(∫  R2|f(x+τ2)g(x−τ2)¯|qdτ)1q.

If we integrate equation (4.13) with respect to d⁢𝒙 , then we get

(4.14) 2π|sin⁡γ||sin⁡γ|2p(∫  R2∫  R2|Wf,gα,β(x,η)|pdηdx)1p≤C(p)(∫  R2∫  R2|f(x+τ2)g(x−τ2)¯|qdτdx)1q.

This equation is the same as

(4.15) 2π|sin⁡γ|1−2p(∫  R2∫  R2|Wf,gα,β(x,η)|pdηdx)1p≤41qC(p)(∫  R2|f(τ)|qdτ)1q(∫  R2|g(x)|qdx)1q.

Furthermore,

(∫  R2∫  R2|Wf,gα,β(x,η)|pdηdx)1p≤41qsin⁡|γ|2p−12πC(p)‖f‖Lq(R2)‖g‖Lq(R2),

and the proof is complete. ∎

4.3. Lieb’s inequality

Lieb’s inequality can be generalized to the coupled Wigner–Ville distribution case. Below, we use sharp Hausdorff–Young inequality mentioned earlier to prove Lieb’s inequality concerning the coupled Wigner–Ville distribution. To this interest, we obtain the following important result.

Theorem 4.3. For two functions f,g∈L2⁢(ℝ2) and 2≤p§lt;∞ , one has

(4.16) (∫  R2∫  R2|Wf,gα,β(x,η)|pdηdx)1p≤41q−1p+2q22π|sin⁡γ|2p−1(2q)−2p‖f‖L2(R2)‖g‖L2(R2),

where 1p+1q=1 .

Proof. From equation (4.13), it follows that

(4.17) (∫  R2|Wf,gα,β(x,η)|pdη)1p≤|sin⁡γ|2p−12πC(p)(∫  R2|f(x+τ2)g(x−τ2)¯|qdτ)1q.

Using the substitution 𝒚=𝒙+𝝉2 , we obtain

(4.18) (∫  R2|Wf,gα,β(x,η)|pdη)1p≤41q|sin⁡γ|2p−12πC(p)(∫  R2|f(y)g(2x−y)¯|qdy)1q,

or

(4.19) ∫  R2|Wf,gα,β(x,η)|pdη≤4pq|sin⁡γ|(2p−1)p(2π)pC(p)p(∫  R2|f(y)g(2x−y)¯|qdy)pq.

We integrate both sides of equation (4.19) with respect to d⁢𝒙 and obtain

(4.20) ∫  R2∫  R2|Wf,gα,β(x,η)|pdηdx≤4pq|sin⁡γ|(2p−1)p(2π)pC(p)p(∫  R2∫  R2|f(y)g(2x−y)¯|qdy)pqdx.

Using the change of the variables, 𝒖=2⁢𝒙 gives

(4.21) (∫  R2∫  R2|Wf,gα,β(x,η)|pdηdx)1p≤41q|sin⁡γ|2p−12πC(p)(∫  R2(∫  R2|f(y)g(u−y)¯|qdy)pqdu4)1p.

Hence,

(4.22) (∫  R2∫  R2|Wf,gα,β(x,η)|pdηdx)1p≤41q−1p|sin⁡γ|2p−12πC(p)(∫  R2(∫  R2|f(y)g(u−y)¯|qdy)pqdu)1pq1q.

Equation (4.22) may be expressed as

(4.23) (∫  R2∫  R2|Wf,gα,β(x,η)|pdηdx)1p≤41q−1p|sin⁡γ|2p−12πC(p)((∫  R2((|f|q∗|g|q)(u))pqdu)1pq)1q.

We write relation (4.23) as

(4.24) (∫  R2∫  R2|Wf,gα,β(x,η)|pdηdx)1p≤41q−1p|sin⁡γ|2p−12πC(p)‖|f|q∗|g|q‖Lpq(R2)1q.

Since f,g∈L2⁢(ℝ2) then |f|q , |g|q∈L2⁢(ℝ2) . Applying the Young inequality for |f|q and |g|q with triple (p′,p′,t)=(2q,2q,pq) , we obtain

(4.25) ∥|f|q∗|g|q∥Lt⁢(ℝ2)≤C⁢(p)4q⁢C⁢(t)2q⁢∥|f|q∥Lp′⁢(ℝ2)⁢∥|g|q∥Lp′⁢(ℝ2).

Observe that

(4.26) C⁢(p)⁢C⁢(p)4r⁢C⁢(t)2r=(2q1+1q-1p)⁢(12)q-pp⁢q=4q2⁢(2q)-2p.

Substituting equation (4.26) into equation (4.24) gives

(∫  R2∫  R2|Wf,gα,β(x,η)|pdηdx)1p≤41q−1p+22πq2|sin⁡γ|2p−1(2q)−2p‖f‖L2(R2)‖g‖L2(R2),

which finishes the proof.

4.4. Logarithmic Sobolev-type inequality

In this section, we formulate a Sobolev-type inequality for the CFrWVD. To carry our endeavour, we shall provide some basic definitions.

Definition 4.4. Given the operator 𝒟=(∂∂⁡t1,∂∂⁡t2) , the Sobolev space 𝒮⁢(ℝ2) on ℝ2 is defined as

(4.27) S(R2)={f∈L2(R2);Df∈L2(R2)}.

Definition 4.5. For 1 ≤p<∞ and j>0 , the weighted Lebesgue space 𝒲jp⁢(ℝ2) on ℝ2 is defined by

(4.28) 𝒲jp⁢(ℝ2)={f∈Ll⁢o⁢cp⁢(ℝ2):⟨𝒙⟩j⁢f∈Lp⁢(ℝ2)},

where ⟨𝐱⟩=(1+|𝐱|2)12 is the weight function.

Let us derive the following result.

Theorem 4.6. Let two functions f,g∈𝒮⁢(ℝ2)∩𝒲jp⁢(ℝ2) , and the following inequality be satisfied:

(4.29) ‖f‖L2(R2)2∫  R2|g(x−τ2)¯|2ln⁡(1+|τ|22)dτ+4π2∫  R2∫  R2(ln⁡|1sin4⁡γ|+ln⁡|η|2)|Wf,gα,β(x,η)|2dηdx≥4Γ′(1)Γ(1)‖f‖L2(R2)2‖g‖L2(R2)2,

where Γ⁢(⋅) is gamma function.

Proof. With the help of logarithmic Sobolev-type inequality for the two-dimensional Fourier transform, we have

(4.30) ∫  R2|f(τ)|2ln⁡(1+|τ|22)dτ+∫  R2ln⁡|η|2|F{f}(η)|2dη≥Γ′(1)Γ(1)∫R2|f(τ)|2dτ.

Inserting f⁢(𝝉) with hˇf,g⁢(𝒙,𝝉) and then setting 𝜼 with -M⁢𝜼 results in

(4.31) ∫  R2|hˇf,g(x,τ)|2ln⁡(1+|τ|22)dτ+ln⁡|η|2∫  R2|F{hˇf,g(x,τ)}(η)|2dη≥Γ′(1)Γ(1)∫  R2|hˇf,g(x,τ)|2dτ.

Hence,

(4.32) ∫  R2|hˇf,g(x,τ)|2ln⁡(1+|τ|22)dτ+ln⁡|detM|2ln⁡|η|2×∫  R2|F{hˇf,g(x,τ)}(−Mη)|2|detM|dη≥Γ′(1)Γ(1)∫  R2|hˇf,g(x,τ)|2dτ.

On application of equations (3.3) and (3.5) to equation (4.32), we get

(4.33) ∫  R2|f(x+τ2)g(x−τ2)¯e−ia(γ)|τ|2|2ln⁡(1+|τ|22)dτ+∫  R2(ln⁡|1sin4⁡γ|+ln⁡|η|2)|d−1(γ)eia(γ)|η|2Wf,gα,β(x,η)|2(1sin2⁡γ)dη≥Γ′(1)Γ(1)∫  R2|f(x+τ2)g(x−τ2)¯e−ia(γ)|τ|2|2dτ.

Equation (4.33) can be rewritten as

(4.34) ∫  R2|f(x+τ2)g(x−τ2)¯|2ln⁡(1+|τ|22)dτ+|d−1(γ)|2∫  R2(ln⁡|1sin4⁡γ|+ln⁡|η|2)|Wf,gα,β(x,η)|2(1sin2⁡γ)dη≥Γ′(1)Γ(1)∫  R2|f(x+τ2)g(x−τ2)¯|2dτ.

This equation is equal to

(4.35) ∫  R2|f(x+τ2)g(x−τ2)¯|2ln⁡(1+|τ|22)dτ+4π2sin2⁡γ∫  R2(ln⁡|1sin4⁡γ|+ln⁡|η|2)|Wf,gα,β(x,η)|2(1sin2⁡γ)dη≥Γ′(1)Γ(1)∫  R2|f(x+τ2)g(x−τ2)¯|2dτ.

Integrating both sides of equation (4.35) with respect to d⁢𝒙 , we have

(4.36) ∫  R2∫R2|f(x+τ2)g(x−τ2)¯|2ln⁡(1+|τ|22)dxdτ+4π2sin2⁡γ∫  R2∫  R2ln⁡|1sin4⁡γ||Wf,gα,β(x,η)|2(1sin2⁡γ)dηdx+4π2sin2⁡γ∫  R2∫  R2ln⁡|η|2|Wf,gα,β(x,η)|2(1sin2⁡γ)dηdx≥Γ′(1)Γ(1)∫  R2∫  R2|f(x+τ2)g(x−τ2)¯|2dxdτ.

Equation (4.36) can be expressed as

(4.37) ∫  R2|f(x+τ2)|2dx∫  R2|g(x−τ2)¯|2ln⁡(1+|τ|22)dτ+4π2sin2⁡γ∫  R2∫  R2ln⁡|1sin4⁡γ||Wf,gα,β(x,η)|2(1sin2⁡γ)dηdx+4π2sin2⁡γ∫  R2∫  R2ln⁡|η|2|Wf,gα,β(x,η)|2(1sin2⁡γ)dηdx≥Γ′(1)Γ(1)∫  R2|f(x+τ2)|2dx∫  R2|g(x−τ2)¯|2dτ.

Equation (4.37) can be rewritten in the form

‖f‖L2(  R2)2∫R2|g(x−τ2)¯|2ln⁡(1+|  τ|22)dτ+4π2∫  R2∫  R2ln⁡|1sin4⁡γ||Wf,gα,β(x,η)|2dηdx+4π2∫  R2∫  R2ln⁡|η|2|Wf,gα,β(x,η)|2dηdx≥4Γ′(1)Γ(1)‖f‖L2(R2)2‖g‖L2(R2)2.

This equation is equal to

‖f‖L2(R2)2∫R2|g(x−τ2)¯|2ln⁡(1+|τ|22)dτ+4π2∫  R2∫  R2(ln⁡|1sin4⁡γ|+ln⁡|η|2)|Wf,gα,β(x,η)|2dηdx≥4Γ′(1)Γ(1)‖f‖L2(R2)2‖g‖L2(R2)2,

which completes the proof.

Remark 4.7. The authors of [15] have presented several uncertainty principles related to CFrWVD such as Hardy’s and Beurling’s uncertainty inequalities which were not investigated in this article. The proof of their uncertainty principles used the definition of the CFrWVD and its properties, while our work is derived by developing the basic relationship between the CFrWVD and the Fourier transform.

5. Conclusion

In this article, we have introduced the CFrWVD and investigated its properties. Also, we presented the close link between the CFrWVD and the Fourier transform. We combined this relation and properties of the CFrWVD to see for several versions of the uncertainty principles related to the proposed transformation. The uncertainty inequalities play a key role in understanding and development of signal analysis.

Ethics

This work did not require ethical approval from a human subject or animal welfare committee.

Data accessibility

This article has no additional data.

Declaration of AI use

We have not used AI-assisted technologies in creating this article.

Authors’ contributions

A.T.A.N.: conceptualization; M.B.: formal analysis; N.B.: investigation; A.R.: validation.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

We declare we have no competing interests.

Funding

The work was supported in part by a grant from Ministry of Research, Technology and Higher Education, Indonesia under WCR (World Class Research) scheme 2023.
==== Refs
References

1. Li BZ , Xu TZ . 2012 Parseval relationship of samples in the fractional Fourier fransform domain. J. Appl. Math. 2012 , 428142. (10.1155/2012/428142)
2. Ozaktas HM , Zalevsky Z , Kutay MA . The fractional Fourier transform with applications in optics and signal processing. New York, NY: Wiley.
3. Zayed A . 2018 Two-dimensional fractional Fourier transform and some of its properties. Integral Transforms Spec. Funct. 29 , 553–570. (10.1080/10652469.2018.1471689)
4. Kamalakkannan R , Roopkumar R . 2019 Multidimensional fractional Fourier transform and generalized fractional convolution. Integral Transforms Spec. Funct. 31 , 152–165. (10.1080/10652469.2019.1684486)
5. Bahri M , Karim SAA . 2023 Fractional fourier transform: main properties and inequalities. Mathematics 11 , 1234. (10.3390/math11051234)
6. Almeida LB . 1994 The fractional Fourier transform and time-frequency representations. IEEE Trans. Signal Process. 42 , 3084–3091. (10.1109/78.330368)
7. Zayed ZI . 1998 A convolution and product theorem for the fractional fourier transform. IEEE Signal Process. Lett. 5 , 101–103. (10.1109/97.664179)
8. Kaur N , Gupta B , Verma AK . 2023 Multidimensional fractional wavelet transforms and uncertainty principles. J. Comput. Appl. Math. 430 , 115250. (10.1016/j.cam.2023.115250)
9. Namias V . 1980 The fractional order Fourier transform and its application to quantum mechanics. IMA J. Appl. Math. 25 , 241–265. (10.1093/imamat/25.3.241)
10. Shah FA , Lone WZ , Nisar KS , Abdeljawad T . 2022 On the class of uncertainty inequalities for the coupled fractional Fourier transform. J. Inequal. Appl. 2022 , 133. (10.1186/s13660-022-02873-2)
11. Kamalakkannan R , Roopkumar R , Zayed A . 2021 On the extension of the coupled fractional Fourier transform and its properties. Integral Transform Spec. Funct. 33 , 65–80. (10.1080/10652469.2021.1902320)
12. Kamalakkannan R , Roopkumar R , Zayed A . 2021 Short time coupled fractional Fourier transform and the uncertainty principle. Fract. Calc. Appl. Anal. 24 , 667–688. (10.1515/fca-2021-0029)
13. Teali AA , Shah FA , Tantary AY , Nisar KS . 2023 Coupled fractional Wigner distribution with applications to LFM signals. Fractals 31 , 2340020. (10.1142/S0218348X23400200)
14. Tao R , Li YL , Wang Y . Short-time fractional Fourier transform and its applications. IEEE Trans. Signal Process. 58 , 2568–2580. (10.1109/TSP.2009.2028095)
15. Firdous FA , Aajaz AT . 2023 Uncertainty principles for the coupled fractional Wigner distribution. Int. J. Geom. Methods Mod. Phys. 20 , 2350017. (10.1142/S0219887823500172)
16. Debnath L , Shah FA . 2015 Wavelet transforms and their applications. Boston, MA: Birkhäuser. (10.1007/978-0-8176-8418-1)
17. Gröchenig K . 2011 Foundation of time-frequency analysis. Boston, MA: Birkhäuser.
18. Bracewell R . 2000 The Fourier transform and its applications. New York, NY: McGraw Hill.
19. Bahri M , Ashino R . 2016 Some properties of windowed linear canonical transform and its logarithmic uncertainty principle. Int. J. Wavelets. Multiresolut. Inf. Process. 14 , 1650015. (10.1142/S0219691316500156)
20. Bahri M . 2022 Windowed linear canonical transform: its relation to windowed Fourier transform and uncertainty principles. J. Inequal. Appl. 2022 , 4. (10.1186/s13660-021-02737-1)
21. Guanlei X , Xiaotong W , Xiaogang X . 2009 The logarithmic, Heisenberg’s and short-time uncertainty principles associated with fractional Fourier transform. Signal Process. 89 , 339–343. (10.1016/j.sigpro.2008.09.002)
22. Gao WB , Li BZ . 2021 Uncertainty principles for the short-time linear canonical transform of complex signals. (https://arxiv.org/abs/1910.00499)
23. Topan A , Bahri M , Bachtiar N , Rangkuti A , Nur M . 2023 Two new relations for the coupled fractional fourier transform. J. Southwest Jiaotong Univ. 58 , 354–358. (10.35741/issn.0258-2724.58.2.32)
24. Shi J , Liu X , Zhang N . 2012 On uncertainty principle for signal concentrations with fractional Fourier transform. Signal Process. 92 , 2830–2836. (10.1016/j.sigpro.2012.04.008)
25. Shah FA , Nisar KS , Lone WZ , Tantary AY . 2021 Uncertainty principles for the quadratic‐phase Fourier transforms. Math. Methods Appl. Sci. 44 , 10416–10431. (10.1002/mma.7417)
