Stationary phase with Cauchy singularity. A critical point of signature $(+,-)$

This paper presents asymptotic expressions for a solid Cauchy transform with a rapidly oscillating phase and a Cauchy singularity, utilizing Stokes' theorem to decompose the integral into three terms that are analyzed via special functions on steepest descent contours in $\mathbb{C}^2$.

The Gist

The Big Picture: Finding the "Sweet Spot" in a Noisy Room

Imagine you are trying to listen to a specific sound (a stationary point) in a very loud, chaotic room. This sound is part of a complex mathematical formula used to solve problems in physics, like how waves move in water or how electricity flows through a body (Electrical Impedance Tomography).

The formula involves an integral, which is essentially a way of adding up millions of tiny contributions to find a total result. The challenge is that the formula has two "troublemakers":

1. The Stationary Point: A place where the wave pattern is smooth and predictable (like a calm spot in a storm).
2. The Singularity (The Pole): A place where the formula blows up or becomes infinite (like a sudden, deafening scream).

Usually, mathematicians have a standard toolkit to handle these troublemakers if they are far apart. But this paper tackles the difficult scenario where the Stationary Point and the Singularity are practically hugging each other. When they are this close, the standard tools break down.

The Problem: When the Map Fails

The authors are studying a specific type of integral that depends on a tiny number, $h$ (think of $h$ as the "grain size" of reality; the smaller it is, the more detailed and wiggly the waves become).

• The Easy Case: If the "scream" (singularity) is far away from the "calm spot" (stationary point), you can use standard techniques (like the "Method of Steepest Descent") to approximate the answer. It's like listening to a conversation in a quiet room; you can easily ignore the noise.
• The Hard Case: If the scream is right next to the calm spot, the standard methods fail. The waves oscillate so wildly that you can't just pick one path to follow.

The Solution: A New Way of Looking at the Room

To solve this, the authors use a clever trick called Polarization.

The Analogy: The Shadow Puppet Trick
Imagine you are trying to understand a 2D shadow on a wall, but the shadow is too messy to analyze directly. Instead of staring at the wall, you step back and realize the shadow is cast by a 3D object. By treating the shadow as a slice of a 3D object, you gain a new perspective.

In the paper, the authors take their 2D problem (the complex plane) and "lift" it into a 4D space (specifically, a 2D slice of a 4D space called $\mathbb{C}^2$). They treat the variable $\omega$ and its "partner" $\bar{\omega}$ as two separate, independent variables.

Once they are in this higher-dimensional space, they can draw new paths (contours) that the calculation can follow. It's like finding a secret tunnel that bypasses the traffic jam.

The Three-Part Breakdown

Using a powerful mathematical tool called Stokes' Theorem (which is like a generalized version of the "Fundamental Theorem of Calculus" for shapes), they slice the messy integral into three distinct pieces:

1. Term I (The "Gaussian" Part):
   This part captures the behavior right where the stationary point and the singularity are interacting. The authors show that this piece can be described using special mathematical functions (related to Dawson's integral, which describes how particles diffuse). Think of this as the "core" of the problem, which they successfully mapped out.

2. Term II (The "Boundary" Part):
   This part comes from the edge of the region they are studying. It turns out this piece is also calculable and gives a specific, predictable value depending on the direction the singularity is facing. It's like the "echo" bouncing off the walls of the room.

3. Term III (The "Noise" Part):
   This is the leftover piece. The authors prove that as the tiny number $h$ gets smaller, this piece becomes vanishingly small (mathematically, it goes to zero faster than any power of $h$). It's the background static that you can safely ignore.

The Result: A New Formula

By combining these three pieces, the authors provide a new asymptotic formula.

• What it means: They have created a "cheat sheet" that tells you exactly what the answer will be when the stationary point and the singularity are very close, without needing to run a supercomputer to simulate every single wave.
• The "Signature": The paper specifically focuses on a case where the wave shape looks like a saddle (up in one direction, down in the other), which is a common shape in physics.

Why This Matters (According to the Paper)

The paper mentions that these integrals appear in:

• Davey-Stewartson Equations: Mathematical models for water waves in two dimensions.
• Electrical Impedance Tomography (EIT): A medical imaging technique that uses electricity to see inside the body (like a CT scan but without radiation).
• Random Matrix Theory: Used in statistics and physics to understand complex systems.

The authors state that their work is the first step in extending these calculations to more complex functions found in these real-world applications. They aren't solving the medical scan or the water wave problem directly in this paper; they are providing the precise mathematical "lens" needed to see the solution clearly when the standard tools are too blurry.

Summary in One Sentence

The authors developed a new mathematical "lens" (using higher-dimensional geometry and contour deformation) to accurately calculate complex wave integrals when a smooth wave pattern and a sudden mathematical singularity are dangerously close to each other, breaking the problem into three manageable parts and proving that the messy leftovers disappear.

Technical Summary

Technical Summary: Stationary Phase with Cauchy Singularity

Problem Statement
This paper investigates the asymptotic behavior of the integral
$$I(a, \phi, \zeta; h) = -\frac{1}{\pi} \int_{\mathbb{C}} \frac{1}{\omega - \zeta} a(\omega; h) e^{\frac{i}{h}\phi(\omega)} L(d\omega)$$
as the small parameter $h \to 0$. The integral represents a solid Cauchy transform of a function with a rapidly oscillating phase $\phi$ and a small parameter $h$. The amplitude $a$ is a classical symbol, and the phase $\phi$ possesses a finite number of non-degenerate stationary points $\omega_k$. The singularity of the integrand is located at $\zeta \in \mathbb{C}$.

The primary challenge addressed is the regime where the singularity $\zeta$ is close to a stationary point of the phase, specifically when $|\zeta - \omega_k| = O(\sqrt{h})$. In this "near-field" regime, standard steepest descent methods fail because the stationary point and the pole cannot be separated. This problem arises naturally in the asymptotic analysis of $d$-bar problems associated with integrable equations in 2D (such as the Davey-Stewartson II equation) and in electrical impedance tomography (EIT). While previous works [13, 14] provided asymptotic formulae for cases where the stationary point is far from the singularity or for specific compact domains, this paper aims to extend these results to more general functions $a$ and $\phi$ appearing in the reflection coefficient of the Davey-Stewartson system.

Methodology
The authors employ a polarization approach combined with Stokes' theorem on complex contours.

1. Polarization and Complexification: The real plane $\mathbb{R}^2_{\xi, \eta}$ is identified with the anti-diagonal in $\mathbb{C}^2_{\omega, \tilde{\omega}}$ via the relations $\omega = \xi + i\eta$ and $\tilde{\omega} = \xi - i\eta$. The integral is lifted to $\mathbb{C}^2$, treating $\omega$ and $\tilde{\omega}$ as independent variables initially, with the constraint $\tilde{\omega} = \omega$ defining the original real domain.
2. Contour Deformation: A smooth family of 2-dimensional contours $\Gamma_\theta$ in $\mathbb{C}^2$ is constructed, interpolating between the anti-diagonal ($\theta=0$) and a Cartesian product of steepest descent lines ($\theta=1$). Specifically, $\Gamma_1 = e^{i\pi/4}\mathbb{R} \times e^{i\pi/4}\mathbb{R}$.
3. Stokes' Decomposition: By applying Stokes' theorem to the deformation of the integration contour from $\Gamma_0$ to $\Gamma_{1-\delta}$, the original integral is decomposed into three distinct terms:
   $$I(a\chi, \phi, \zeta; h) = I(\zeta, 1) + II(\zeta, 1) + III(\zeta, 1)$$
   (in the limit $\delta \to 0$).
   • Term I: An integral over the Cartesian product contour $\Gamma_1$.
   • Term II: An integral arising from the singularity crossing the deformation path (related to the residue at $\zeta$).
   • Term III: An integral over the "side" of the deformation tube, involving the derivative of the cutoff function.

Key Contributions and Results

The paper provides detailed asymptotic expansions for each of the three terms under the assumption that the Hessian of the phase at the stationary point (assumed to be at the origin) has signature $(+, -)$.

1. Direct Approach (Far Field): In Section 2, the authors confirm that when $|\zeta| \gg \sqrt{h}$, standard stationary phase methods apply, yielding a sum of a contribution from the stationary point and a contribution from the pole, with the latter being of the form $c(\zeta; h)e^{i\phi(\zeta)/h}$.

2. Analysis of Term I (Section 4):

   • This term is analyzed using the method of steepest descent on the Cartesian product contour $\Gamma_1$.
   • The inner integral with respect to $\tilde{\omega}$ is evaluated using the stationary phase method, reducing the problem to a 1D integral over $\omega$.
   • The leading order behavior is expressed in terms of special functions $G_{\rho}^{l/r}$, which are related to the Hilbert transform of the Dawson integral.
   • Result: The leading term involves a scaling factor $h^{1/2}$ and the special function evaluated at a scaled variable $h^{-1/2}\zeta$. The specific branch ($l$ or $r$) depends on the sign of $\text{Im}(e^{-i\pi/4}\zeta)$.

3. Analysis of Term II (Section 5):

   • This term captures the interaction between the pole and the phase. The authors analyze this for intermediate values of $\varepsilon = |\zeta|/\sqrt{h}$ and the small $\varepsilon$ limit ($\varepsilon \le h^\delta$).
   • For small $\varepsilon$, the integral is expanded in powers of $\varepsilon$ and $h$. The expansion involves polynomials in $\varepsilon$, logarithmic terms $\ln(1/\varepsilon)$, and powers of $h$.
   • Result: The leading order contribution of Term II is a constant (independent of $\zeta$ in the leading order scaling) multiplied by $h^{1/2}$, with a coefficient that depends on the sign of $\text{Im}(e^{-i\pi/4}\zeta)$. Specifically, it contributes $\pm \frac{1}{2}$ times the leading term of Term I.

4. Analysis of Term III (Section 6):

   • The authors prove that this term is exponentially small, specifically $O(h^\infty)$, provided the cutoff function $\chi$ has sufficiently small support near the origin. This validates the decomposition by showing the error introduced by the deformation is negligible.

5. Main Theorem (Theorem 1.2):

   • Combining the results for $I$, $II$, and $III$, the paper presents the leading-order asymptotic formula for the integral when $|\zeta| \in ]0, h^{1/2-\delta}[$.
   • The result is a unified expression involving the special functions $G_{-\omega^2/2}^{l/r}$ and a step-function-like correction term ($\pm 1/2$) that accounts for the relative position of $\zeta$ with respect to the steepest descent path.

Significance and Claims
The paper claims to provide the first step in extending asymptotic methods for $d$-bar problems to more general functions $a$ and $\phi$, specifically those arising in the reflection coefficient of the Davey-Stewartson II equation. The significance lies in:

• Handling the Critical Regime: Providing a rigorous asymptotic framework for the case where the singularity $\zeta$ is within $O(\sqrt{h})$ of the stationary point, a regime where standard numerical methods fail due to high oscillation and where standard steepest descent is inapplicable.
• Polarization Technique: Demonstrating the utility of the polarization method (lifting to $\mathbb{C}^2$) combined with Stokes' theorem to decompose the integral into manageable parts.
• Special Functions: Expressing the solution in terms of specific transcendental functions (related to Dawson's integral) that capture the transition behavior near the critical point.

The authors note that the current study is restricted to the signature $(+, -)$ of the Hessian. They acknowledge that other signatures and degenerate cases are important for applications (as seen in numerical studies of the DS II equation) but state that the application of this approach to those cases will be the subject of future research. The paper does not propose new numerical algorithms but rather provides the theoretical asymptotic formulae necessary for hybrid numerical-asymptotic schemes.