Airy Transform: Theory & Applications

• Airy Transform is a linear integral transform defined by the Airy function kernel, exhibiting third-order oscillatory behavior and convolutional structure.
• It features unitary involution on L²(ℝ) and encodes operator mappings that preserve the Weyl canonical commutation relations and bispectrality.
• The transform underpins applications in random matrix theory and optics, enabling analysis of Tracy–Widom distributions and non-diffracting Airy beams.

The Airy transform is a linear integral transform with kernel given by the Airy function $\mathrm{Ai}(x)$, which arises intrinsically in the analysis of partial differential equations with linear or higher-order polynomial potentials, diffraction phenomena in optics, the propagation of non-spreading wave packets, and the spectral theory of integral operators in random matrix theory. Central properties of the Airy transform include its convolutional structure, its unitary involutive behavior on $L^2(\mathbb{R})$, rich bispectrality, connections to higher-order heat equations, and applications to physical systems displaying third-order dispersion. In several applied and theoretical contexts, the Airy transform unifies distinct mathematical frameworks and reveals deep structural correspondences between operator theory, special functions, and physical wave phenomena (Shen et al., 2021, Babusci et al., 2010, Dattoli et al., 2010).

1. Formal Definitions and Operator Structure

For a sufficiently well-behaved function $f(x)$, the full-line Airy transform $\mathcal{A}$ is defined by

$$(\mathcal{A}f)(y) = \int_{-\infty}^{\infty} \mathrm{Ai}(y - a x)\, f(x)\, dx$$

for a scaling parameter $a > 0$, with the case $a=1$ often used for simplicity. The kernel $\mathrm{Ai}(y - a x)$ encodes third-order oscillations. The transform is invertible with the same kernel: $$f(x) = \int_{-\infty}^{\infty} (\mathcal{A}f)(\eta)\, \mathrm{Ai}(a x - \eta)\, d\eta$$ (Dattoli et al., 2010). On $L^2(\mathbb{R})$, $\mathcal{A}$ is a unitary involution, satisfying $\mathcal{A}^2 = I$.

Operator mappings under the Airy transform are as follows: $$\mathcal{A}_a\{x\, f(x)\}(\eta) = (\eta - a^3\, \partial_\eta^2)\, (\mathcal{A}_a f)(\eta), \qquad \mathcal{A}_a\{\partial_x f(x)\}(\eta) = \partial_\eta (\mathcal{A}_a f)(\eta)$$ These maps encode the Weyl canonical commutation relations: $[\eta, \,\partial_\eta] = 1$ is preserved (Dattoli et al., 2010, Babusci et al., 2010).

A related operator realization expresses Airy evolution via the exponential of the third derivative: $F(x, y) = e^{y\, \partial_x^3} f(x)$ which is the unique solution to $\partial_y F(x, y) = \partial_x^3 F(x, y)$, $F(x, 0) = f(x)$ (Babusci et al., 2010).

2. Integral and Spectral Representations

The Airy transform possesses various integral representations:

• Fourier representation of Airy function:

$$\mathrm{Ai}(t) = \frac{1}{2\pi} \int_{-\infty}^\infty e^{i\, (t\,\xi + \xi^3/3)}\, d\xi$$

• Exponential–Airy convolution identity:

$$e^{\lambda x^3} = \int_{-\infty}^{\infty} e^{\lambda^{1/3} x t}\, \mathrm{Ai}(t)\, dt$$

• Airy transform kernel (parameterized by diffusion time $y$):

$$(A_y f)(x) = \frac{1}{\sqrt[3]{3y}} \int_{-\infty}^{\infty} \mathrm{Ai}\Bigl(-\frac{x-\xi}{\sqrt[3]{3y}}\Bigr) f(\xi)\, d\xi$$

where $y>0$ (Babusci et al., 2010).

The unitarity and involutive property $\mathcal{A}^2 = I$ on $L^2(\mathbb{R})$ stem from the symmetry and completeness of the Airy function on the real line (Shen et al., 2021).

3. Airy Integral Operator, Eigendecomposition, and Bispectrality

The Airy integral operator $T:L^2(0, \infty) \to L^2(0, \infty)$ is defined as

$$T[f](x) = \int_0^\infty \mathrm{Ai}(x + y)\, f(y)\, dy$$

Its spectrum consists of a strictly decreasing sequence of positive, simple eigenvalues $\lambda_n \to 0$: $$T[\psi_n](x) = \lambda_n \psi_n(x), \qquad \|\psi_n\|_2 = 1, \quad \psi_n(0)>0$$ Crucially, $T$ commutes with the singular Sturm–Liouville operator

$$L[f](x) = - \frac{d}{dx}\Bigl(x \frac{d}{dx} f(x)\Bigr) + x^2 f(x)$$

They share a complete set of real-valued eigenfunctions, demonstrating bispectrality. The eigenvalue equations are connected via inner products: $\chi_n \psi_n(0) + \psi_n'(0) = 0$, and the eigenvalue $\lambda_n$ is given as a kernel ratio independent of $x$ (Shen et al., 2021).

Asymptotically, eigenvalues decay as $\exp(-c\, n^{3/2})$, while eigenfunctions $\psi_n(x)$ decay superexponentially as $\mathrm{Ai}(x)$ for large $x$.

4. Airy Polynomials, Hermite Polynomials, and Orthogonality

The "Airy polynomials" $\alpha i_n(x, y)$, defined as the Airy transform of monomials,

$$\alpha i_n(x, y) = \frac{1}{\sqrt[3]{3y}} \int_{-\infty}^{\infty} \mathrm{Ai}\Bigl(-\frac{x-\xi}{\sqrt[3]{3y}}\Bigr) \xi^n\, d\xi$$

are proven to coincide with the third-order Hermite polynomials $H_n^{(3)}(x, y) = e^{y\partial_x^3} x^n$ (Babusci et al., 2010).

These polynomials satisfy recursions: $$\frac{\partial}{\partial x} H_n^{(3)}(x,y) = n\, H_{n-1}^{(3)}(x,y),\quad H_{n+1}^{(3)}(x,y) = x\, H_n^{(3)}(x,y) + 3 \frac{n!}{(n-2)!} H_{n-2}^{(3)}(x,y)$$ and the differential equation

$$\Bigl(y\, \frac{d^3}{dx^3} + x\, \frac{d}{dx}\Bigr) H_n^{(3)}(x, y) = n\, H_n^{(3)}(x, y)$$

The expansion $f(x) = \sum_{n=0}^\infty a_n H_n^{(3)}(x, -|y|)$ admits explicit coefficients involving derivatives of $\mathrm{Ai}(x)$, with an implicit orthogonality measure given by $\mathrm{Ai}(x/\sqrt[3]{3|y|})$ (Babusci et al., 2010).

5. Airy Transform Uncertainty Principle

For $f \in L^2(\mathbb{R})$ with Airy transform $\sigma(x) = \mathcal{A} f(x)$ supported on $[a, \infty)$, the fraction of $L^2$ norm localized to $[b, \infty)$ satisfies

$$\alpha^2 = \frac{\int_b^\infty |f(x)|^2 dx}{\int_{-\infty}^{\infty} |f(x)|^2 dx} \leq \int_b^\infty \int_a^\infty \mathrm{Ai}(x+y)^2\, dy\, dx$$

The sharp bound is attained by maximizing the leading singular value $\lambda_0(a+b)$ of the truncated Airy operator. This expresses a nontrivial uncertainty relation for simultaneous localization of $f$ and its Airy transform—analogous in spirit to the Fourier uncertainty principle but reflecting the non-symmetric, oscillatory kernel of the Airy transform (Shen et al., 2021).

6. Applications in Random Matrix Theory and Optics

The Airy transform and its integral operator play key roles in two major domains:

Domain	Airy Transform Role	Reference
Random matrix theory	Fredholm determinant of Airy kernel integral operators yields Tracy–Widom distributions at soft edge of GUE; eigenvalues of $T$ enter explicitly in distributional computations.	(Shen et al., 2021)
Optics	Governs propagation of finite-energy Airy beams under paraxial wave equation. The transform is used to derive maximally concentrated, non-diffracting beams.	(Shen et al., 2021, Babusci et al., 2010)

In random matrix theory, the Airy kernel $$K_{Ai}(x,y) = \int_s^\infty \mathrm{Ai}(x+z-s)\mathrm{Ai}(y+z-s)\, dz$$ acting on $L^2[s,\infty)$ is decomposed into Airy transform operations. This framework permits numerically stable, relative-precision computation of Tracy–Widom distributions (Shen et al., 2021).

In optics, the Airy transform is central to the design of Airy beams and the analysis of diffraction catastrophes. The optimal finite-energy Airy beam initial profile is provided by the leading eigenfunction of the truncated Airy integral operator. The convolution structure of the Airy transform underpins the solution of the paraxial wave and Schrödinger equations with linear potentials, yielding explicitly non-spreading Airy wave packets (Dattoli et al., 2010, Babusci et al., 2010).

7. Connections and Generalizations

The Airy transform and its associated operators provide a template for broader families of integral transforms:

• Generalized higher-order Airy transforms arise by replacing the third derivative in $\exp( y \partial_x^3 )$ with higher-odd-order derivatives, yielding solutions of corresponding higher-order heat equations. These generalizations align with Watson's higher-degree Airy-type functions and connect to Bessel expansions (Babusci et al., 2010).
• The analogy between the Airy transform and the Gauss–Weierstrass (heat) transform highlights their roles in diagonalizing third-order and second-order dispersive operators, respectively. Unlike the diffusive Gaussian kernel, the Airy kernel encodes oscillatory, non-dispersive behavior (Dattoli et al., 2010).

Outstanding questions and conjectures concern the development of explicit orthogonality theories for Airy (Hermite-type) polynomials, combinatorial interpretations, and further spectral-theoretic characterizations (including Bell-type identities and uncertainty relations) (Babusci et al., 2010, Shen et al., 2021).

---

References:

(Shen et al., 2021): On the Evaluation of the Eigendecomposition of the Airy Integral Operator (Babusci et al., 2010): The Airy transform and the associated polynomials (Dattoli et al., 2010): Linear Potentials, Airy Wave Packets and Airy Transform