Stochastic Airy Function in Random Matrix Theory

• The stochastic Airy function is a random eigenfunction and entire solution that encapsulates soft-edge scaling limits in β-ensembles and random matrix theory.
• It is rigorously defined through operator-theoretic and quasi-derivative formulations involving Brownian motion, leading to a unique square-integrable solution with discrete spectrum.
• Its various representations, including semigroup limits and Feynman–Kac formulas, connect it to Tracy–Widom laws and higher-order soft-edge universality.

Searching arXiv for recent and foundational papers on the stochastic Airy function and related operator/semigroup formulations. In random matrix theory and random Sturm–Liouville theory, the stochastic Airy function is the eigenfunction-level object associated with the stochastic Airy operator on the half-line, and, in a more refined analytic formulation, the unique up-to-scaling square-integrable solution of the stochastic Airy equation. It governs soft-edge scaling limits for $\beta$-ensembles, identifies the $\beta$-Tracy–Widom spectrum through its boundary zeros, and admits semigroup and Feynman–Kac representations involving Brownian motion, Brownian bridges, or reflected Brownian motions with local times [(Krishnapur et al., 2013); (Minami, 2014); (Gorin et al., 2016); (Lamarre et al., 2017); (Lambert et al., 2020)]. The term is not completely uniform across the literature: a separate fractional-PDE line uses it for the fractional Airy function $\Ai_\alpha$ and its stochastic expectation representation (Marchione et al., 2022).

1. Operator-theoretic definition

For fixed $\beta>0$, let $W(x)$, $x\ge 0$, be a standard one-dimensional Brownian motion, and write $B'(x)=dW/dx$ in the sense of distributions. The stochastic Airy operator is the random differential operator

$$L_\beta=-\frac{d^2}{dx^2}+x+\frac{2}{\sqrt{\beta}}\,B'(x)$$

acting on $[0,\infty)$ with Dirichlet boundary condition at $0$ (Krishnapur et al., 2013).

A precise realization is obtained from the quadratic form

$\beta$0

with form domain

$\beta$1

Using Kato–Rellich and form-boundedness of the stochastic term, $\beta$2 is almost surely bounded below and closed, hence defines a unique self-adjoint operator (Krishnapur et al., 2013).

Minami gives an equivalent generalized Sturm–Liouville construction. Writing $\beta$3 with $\beta$4 and introducing the quasi-derivative

$\beta$5

one defines the formal expression

$\beta$6

where $\beta$7 is Gaussian white noise, on a class of functions for which both $\beta$8 and $\beta$9 are absolutely continuous and the associated quasi-derivative expression belongs to $\Ai_\alpha$0. On $\Ai_\alpha$1 with $\Ai_\alpha$2, the resulting limit-point restriction at $\Ai_\alpha$3 is almost surely self-adjoint and has purely discrete spectrum (Minami, 2014).

The spectral picture is correspondingly rigid. Almost surely, the spectrum is bounded below, simple, and discrete, with eigenpairs

$\Ai_\alpha$4

and the eigenfunctions form a complete orthonormal system in $\Ai_\alpha$5 (Krishnapur et al., 2013).

2. Eigenfunctions, regularity, and Riccati structure

The phrase “stochastic Airy functions” most directly refers to these eigenfunctions $\Ai_\alpha$6. They are real and continuous, satisfy

$\Ai_\alpha$7

and form an almost surely complete orthonormal basis of $\Ai_\alpha$8 (Krishnapur et al., 2013). On each compact interval, $\Ai_\alpha$9 is almost surely continuously differentiable, and the family depends measurably on the Brownian path; for each fixed $\beta>0$0, the map from the Brownian path to $\beta>0$1 is continuous in uniform norm on compact sets (Krishnapur et al., 2013).

In Minami’s quasi-derivative formulation, a stochastic Airy function at spectral parameter $\beta>0$2 is a nontrivial solution of

$\beta>0$3

equivalently

$\beta>0$4

with $\beta>0$5 (Minami, 2014). This can be rewritten as the first-order system

$\beta>0$6

or, formally,

$\beta>0$7

with the differential equation understood in the distributional or quasi-derivative sense (Minami, 2014).

For each fixed realization and $\beta>0$8, Minami constructs two distinguished solutions: $\beta>0$9, characterized by $W(x)$0 and $W(x)$1, and $W(x)$2, the principal solution square-integrable at $W(x)$3 (Minami, 2014). Standard ODE and Volterra arguments show that these solutions are jointly continuous in $W(x)$4 and measurable in the randomness. The principal solution obeys the almost-sure asymptotic estimate

$W(x)$5

so the random white-noise potential remains subdominant to the linear confining term in the leading decay (Minami, 2014).

The Riccati transform provides a probabilistic encoding of zeros and eigenvalue index. If

$W(x)$6

then

$W(x)$7

and for $W(x)$8 the process explodes to $W(x)$9 exactly $x\ge 0$0 times (Minami, 2014). The same paper states the nodal-counting theorem: the $x\ge 0$1-th eigenfunction has exactly $x\ge 0$2 zeros in $x\ge 0$3.

3. The random entire function $x\ge 0$4

A second, more analytic usage appears in the study of characteristic polynomials of the Gaussian $x\ge 0$5-ensemble. Here the stochastic Airy function is a random entire function $x\ge 0$6, defined as the unique, up to scaling, $x\ge 0$7 solution on $x\ge 0$8 of the stochastic Airy equation (Lambert et al., 2020).

The equation is a second-order Itô SDE. Let $x\ge 0$9 be a two-sided Brownian motion with

$B'(x)=dW/dx$0

Then for each $B'(x)=dW/dx$1,

$B'(x)=dW/dx$2

Equivalently, with

$B'(x)=dW/dx$3

the pair $B'(x)=dW/dx$4 satisfies a Volterra integral system, and existence and uniqueness follow by Picard iteration (Lambert et al., 2020).

For fixed $B'(x)=dW/dx$5, the Dirichlet and Neumann solutions are defined by

$B'(x)=dW/dx$6

and their Wronskian is identically $B'(x)=dW/dx$7, so they form a fundamental system (Lambert et al., 2020). The square-integrable solution $B'(x)=dW/dx$8 is obtained by taking a large-$B'(x)=dW/dx$9 limit of the Dirichlet solutions with a WKB normalization. The resulting function is $$L_\beta=-\frac{d^2}{dx^2}+x+\frac{2}{\sqrt{\beta}}\,B'(x)$$0 in $$L_\beta=-\frac{d^2}{dx^2}+x+\frac{2}{\sqrt{\beta}}\,B'(x)$$1, entire in $$L_\beta=-\frac{d^2}{dx^2}+x+\frac{2}{\sqrt{\beta}}\,B'(x)$$2, and decays to $$L_\beta=-\frac{d^2}{dx^2}+x+\frac{2}{\sqrt{\beta}}\,B'(x)$$3 as $$L_\beta=-\frac{d^2}{dx^2}+x+\frac{2}{\sqrt{\beta}}\,B'(x)$$4 (Lambert et al., 2020).

Its large-$$L_\beta=-\frac{d^2}{dx^2}+x+\frac{2}{\sqrt{\beta}}\,B'(x)$$5 behavior is Airy-like but random:

$$L_\beta=-\frac{d^2}{dx^2}+x+\frac{2}{\sqrt{\beta}}\,B'(x)$$6

uniformly for $$L_\beta=-\frac{d^2}{dx^2}+x+\frac{2}{\sqrt{\beta}}\,B'(x)$$7 in compact sets, with a corresponding asymptotic for $$L_\beta=-\frac{d^2}{dx^2}+x+\frac{2}{\sqrt{\beta}}\,B'(x)$$8 (Lambert et al., 2020). The same work proves that the zeros of $$L_\beta=-\frac{d^2}{dx^2}+x+\frac{2}{\sqrt{\beta}}\,B'(x)$$9 coincide with the eigenvalues of the stochastic Airy operator $[0,\infty)$0; equivalently, they form the Airy$[0,\infty)$1 point process (Lambert et al., 2020).

This formulation is tied directly to edge asymptotics of Gaussian $[0,\infty)$2-ensemble characteristic polynomials. The rescaled characteristic polynomial converges to the random entire function $[0,\infty)$3 near the spectral edge, and a coupling based on the transfer-matrix recurrence and a KMT embedding yields the quantitative estimate that, for any $[0,\infty)$4, the discrete object and $[0,\infty)$5 are uniformly close by $[0,\infty)$6 with overwhelming probability (Lambert et al., 2020). The same paper records a shift-invariance in law:

$[0,\infty)$7

for every $[0,\infty)$8 (Lambert et al., 2020).

4. Soft-edge universality and the Dyson $[0,\infty)$9-ensemble

The operator and function arise as canonical soft-edge limits of random matrices. Krishnapur, Rider, and Virág study the Jacobi matrix $0$0 associated to the Dyson $0$1-ensemble with uniformly convex polynomial potential $0$2, with entries distributed so that the eigenvalue law is the Coulomb gas

$0$3

They identify explicit local minimizers $0$4 near index $0$5, define the soft-edge quantities

$0$6

and use the scaling

$0$7

to embed the discrete model into $0$8 (Krishnapur et al., 2013).

With the scaled operator

$0$9

one obtains almost-sure norm–resolvent convergence

$\beta$00

and consequently convergence of eigenvalues and eigenvectors: for each fixed $\beta$01, the $\beta$02-th smallest eigenvalue of $\beta$03 converges to the $\beta$04-th eigenvalue of $\beta$05, and the corresponding eigenvectors converge in $\beta$06 (Krishnapur et al., 2013).

On the original random-matrix scale, if $\beta$07 is the $\beta$08-th largest eigenvalue of $\beta$09, then

$\beta$10

and the joint law of $\beta$11 is the $\beta$12-Tracy–Widom law (Krishnapur et al., 2013). This is the sense in which the stochastic Airy function is universal: for any uniformly convex analytic $\beta$13 and any $\beta$14, the top eigenvalues fluctuate according to the same limiting operator, independent of the fine details of the potential (Krishnapur et al., 2013).

The same work conjectures operator limits for nonregular soft edges. If the equilibrium density vanishes like $\beta$15 at the edge, the proposed limiting operator is

$\beta$16

and the scaling exponent is predicted to shift from $\beta$17 to $\beta$18 (Krishnapur et al., 2013). This suggests a hierarchy of “higher-order Tracy–Widom” regimes controlled by stochastic Airy-type operators.

5. Semigroups, path integrals, and spiked extensions

The stochastic Airy function also appears through semigroup limits. Gorin and Shkolnikov analyze high powers of tridiagonal Gaussian $\beta$19-ensemble matrices. If

$\beta$20

with $\beta$21 and $\beta$22, and

$\beta$23

then suitable restricted and scaled powers converge to a random integral operator $\beta$24 on $\beta$25 (Gorin et al., 2016). In the full-space case, the limit is the stochastic Airy semigroup

$\beta$26

where

$\beta$27

with Dirichlet boundary at $\beta$28 (Gorin et al., 2016).

This semigroup has a Feynman–Kac representation. Its kernel is expressed through a Brownian bridge $\beta$29 from $\beta$30 to $\beta$31 over $\beta$32, the bridge local times $\beta$33, and the exponential weight

$\beta$34

with restriction to paths staying in the admissible set $\beta$35 (Gorin et al., 2016). This leads to a probabilistic formula for the Laplace functional of the Airy$\beta$36 point process, obtained by a moment-method to Fredholm-determinant argument (Gorin et al., 2016). A by-product is the Gaussian identity

$\beta$37

for a standard Brownian excursion $\beta$38 and its local times $\beta$39 (Gorin et al., 2016).

Gaudreau Lamarre and Shkolnikov extend this framework to one-spike perturbations. For spike parameter $\beta$40, they define

$\beta$41

where $\beta$42 is reflected Brownian motion on $\beta$43 and $\beta$44 its local time (Lamarre et al., 2017). The resulting semigroup satisfies

$\beta$45

almost surely on $\beta$46, where the spiked stochastic Airy operator is

$\beta$47

subject to the Robin boundary condition

$\beta$48

The case $\beta$49 recovers the non-spiked Dirichlet operator, while $\beta$50 gives the Neumann boundary condition (Lamarre et al., 2017).

The same paper proves a local-time identity for the reflected Brownian bridge. If $\beta$51, $\beta$52, is a reflected Brownian bridge and $\beta$53 denotes its local time, then conditioned on $\beta$54,

$\beta$55

for every $\beta$56 (Lamarre et al., 2017). In the deterministic limit $\beta$57, the white-noise term vanishes and one recovers ordinary Airy boundary-value problems expressed in terms of shifted Airy functions (Lamarre et al., 2017).

6. Terminological variants and related constructions

Within the cited literature, the phrase “stochastic Airy function” appears in several nearby but nonidentical senses.

Usage	Defining object	Reference
Operator eigenfunction	Eigenfunctions of $\beta$58 on $\beta$59	(Krishnapur et al., 2013, Minami, 2014)
Random entire function	Unique $\beta$60 solution $\beta$61 of the stochastic Airy equation	(Lambert et al., 2020)
Fractional-PDE usage	Fractional Airy function $\beta$62 with stochastic expectation representation	(Marchione et al., 2022)

The first two usages are tightly connected. The operator eigenfunctions furnish the spectral basis of the stochastic Airy operator, while the entire-function formulation packages the same soft-edge spectral data into a random holomorphic family whose zeros encode the Airy$\beta$63 process [(Krishnapur et al., 2013); (Lambert et al., 2020)]. A plausible implication is that these are best viewed as two complementary realizations of the same soft-edge object: one spectral and one analytic.

The fractional-PDE usage is structurally different. Marchione and Orsingher define, for $\beta$64,

$\beta$65

with the integer-order specialization

$\beta$66

(Marchione et al., 2022). They relate this function to the space-fractional Cauchy problem

$\beta$67

whose special case $\beta$68 has pseudo-density

$\beta$69

(Marchione et al., 2022).

That paper also gives a stochastic expectation representation. If $\beta$70 is the pseudo-process with density $\beta$71 and $\beta$72 an independent stable subordinator, then for $\beta$73,

$\beta$74

where

$\beta$75

and $\beta$76 is a generalized Gamma random variable (Marchione et al., 2022). The same framework yields a convergent power series for $\beta$77, shows that $\beta$78 is the classical Airy function, proves

$\beta$79

and identifies the time-changed pseudo-process as a stable law with index $\beta$80 and skewness

$\beta$81

with the Cauchy law arising at $\beta$82 (Marchione et al., 2022).

This terminological divergence is a persistent source of confusion. The random-matrix usage concerns eigenfunctions and entire solutions generated by white-noise perturbations of the Airy operator; the fractional-PDE usage concerns integral transforms and stable pseudo-processes (Marchione et al., 2022). The shared terminology reflects formal Airy-type oscillatory structure, but the underlying objects belong to different analytic frameworks.