🔢 mathematics

Boundedness of Fourier Integral Operators with complex phases on Fourier Lebesgue spaces

This paper establishes sharp boundedness estimates for Fourier integral operators with complex phases on Fourier Lebesgue spaces, proving that under a spatial factorization condition of rank $\varkappa$ , such operators are bounded on $\mathcal{F}L^p$ when their order $m$ satisfies $m \leq -\varkappa\left|\frac{1}{p}-\frac{1}{2}\right|$ .

Original authors: Duván Cardona, William Obeng-Denteh, Frederick Opoku

Published 2026-02-17

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Duván Cardona, William Obeng-Denteh, Frederick Opoku

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Picture: Navigating a Foggy Mountain

Imagine you are trying to send a message (a sound wave, a light signal, or a mathematical function) from point A to point B. In the world of physics and mathematics, this is often done using Fourier Integral Operators (FIOs). Think of these operators as high-tech GPS systems that guide the message through a complex landscape.

Usually, these GPS systems work on a clear, flat map (using real numbers). They tell the message exactly where to go. However, sometimes the terrain is so rugged, or the "fog" (mathematical singularities called caustics) is so thick, that a flat map fails. The GPS gets stuck, or the message gets lost.

To fix this, mathematicians invented a new kind of GPS that uses complex numbers (numbers with a "real" part and an "imaginary" part). This is like giving the GPS a 3D view or a thermal camera that can see through the fog. This paper is about proving that this new, high-tech GPS works reliably, even when the terrain is tricky.

The Main Characters

The Operator (The GPS): This is the machine that processes the signal. It takes an input (like a sound wave) and transforms it into an output.
The Phase Function (The Map): This is the rulebook the GPS follows.
- Real Phase: A standard, flat map.
- Complex Phase: A magical map that can bend and twist to avoid obstacles that would trap a standard map.
The Fourier Lebesgue Space (The Terrain): This is the specific type of "ground" the signal is traveling on. Some grounds are smooth (easy to walk on), while others are rocky and uneven (hard to walk on). The paper asks: "How rough can the ground be before the GPS breaks?"

The Problem: When the Map Gets Too Complicated

In the past, mathematicians knew exactly how to handle these GPS systems when the map was simple (real numbers). They knew exactly how much "roughness" (mathematical order) the system could handle before it stopped working.

But when you switch to Complex Phases (the 3D/thermal map), things get messy.

The Oscillation: The complex map makes the signal vibrate and wiggle much more violently than a real map. It's like trying to drive a car on a road that is shaking violently.
The Danger: If the road is too rough (the operator has too high an "order"), the signal might explode or disappear.

The authors wanted to find the Safety Limit. They asked: What is the maximum amount of roughness (order $m$ ) this complex GPS can handle on this specific type of terrain ( $FL^p$ ) without crashing?

The Solution: The "Rank" of the Road

The authors discovered that the safety limit depends on a specific geometric feature of the map, which they call the Rank ( $\kappa$ ).

The Analogy of the Highway:
Imagine the signal is a convoy of trucks.

Rank $\kappa = 0$ : The road is a straight, flat highway. The trucks can go as fast as they want. The system is very stable.
Rank $\kappa = d$ (Maximum): The road is a chaotic, winding mountain pass with no straight lines. The trucks have to slow down significantly to avoid crashing.

The paper proves a specific formula for the Speed Limit (the order $m$ ):
$m \leq -\kappa \times \left| \frac{1}{p} - \frac{1}{2} \right|$

What this means in plain English:

The more "chaotic" the road is (higher $\kappa$ ), the slower the trucks must go (lower $m$ ).
The further you are from the "perfect balance" of the terrain (moving away from $p=2$ ), the slower you must go.
If you try to go faster than this limit, the system breaks. The authors proved this limit is sharp, meaning you cannot push the limit any further; it is the absolute edge of safety.

How They Proved It: The "Slicing" Technique

To prove this, the authors had to deal with the violent shaking (oscillations) caused by the complex map. They used a clever trick called Littlewood-Paley Decomposition.

The Metaphor of the Cake:
Imagine the signal is a giant, messy cake.

Slicing: Instead of trying to eat the whole cake at once, they slice it into tiny, manageable pieces based on frequency (how "fast" the signal is vibrating).
The Angular Decomposition: They then look at each slice from different angles. Because the complex map has a specific structure (the "Spatial Factorization Condition"), they realized that in certain directions, the shaking cancels out, and in others, it doesn't.
The Integration: They calculated how much "energy" remains in each slice after the shaking settles. They found that if the speed limit (order $m$ ) is respected, the total energy of all the slices adds up to a safe, finite number.

Why This Matters

Solving the Unsolvable: This allows mathematicians to solve equations (like the wave equation) in situations where the old "real number" methods fail. It's like finding a way to navigate a storm that was previously considered impossible to cross.
Global Construction: With real maps, you often have to patch together many small maps to cover a large area (like a puzzle). With complex maps, you can often use one single map to cover the whole area. This paper proves that this "one-map" approach is mathematically sound and stable.
Precision: They didn't just say "it works." They gave the exact formula for the limit. This is crucial for engineers and physicists who need to know exactly how much error they can tolerate in their models.

Summary

This paper is a safety manual for a new, advanced type of mathematical GPS.

Old GPS: Works on flat maps (Real numbers).
New GPS: Works on 3D, fog-piercing maps (Complex numbers).
The Discovery: The new GPS is powerful, but it has a strict speed limit depending on how bumpy the road is.
The Result: The authors calculated the exact speed limit and proved that if you stay under it, the system is perfectly stable. If you go over it, it crashes.

This ensures that scientists can use these powerful complex tools to model the universe (like black holes or quantum waves) without fear of the math breaking down.

1. Problem Statement

The paper addresses the boundedness properties of Fourier Integral Operators (FIOs) associated with complex-valued phase functions acting on Fourier Lebesgue spaces ( $FL^p$ ).

Context: While the theory of FIOs with real-valued phases is well-established (notably by Hörmander, Seeger, Sogge, and Stein), operators with complex phases arise naturally in the analysis of wave equations on specific manifolds (e.g., Heisenberg and Métivier groups) and in the construction of parametrices for operators with non-real principal symbols.
The Challenge: Complex phases introduce significant analytical difficulties. Unlike real phases, they do not admit a simple reduction to the real case. Specifically, conjugating the operator with the Fourier transform transforms the symbol into a class with more oscillatory behavior (Hörmander class $S^{m}_{1/2, 1/2}$ rather than the standard Kohn–Nirenberg class $S^{m}_{1,0}$ ), complicating standard $L^p$ and $FL^p$ estimates.
Goal: To establish sharp boundedness estimates for these operators on $FL^p(\mathbb{R}^d)$ spaces, determining the precise order $m$ required for boundedness based on the geometric degeneracy of the phase function.

2. Methodology

The authors employ a combination of microlocal analysis, dyadic decomposition, and complex interpolation. The core methodology involves:

Complex Phase Transformation:
The authors analyze an FIO $T$ with a complex phase $\Phi$ . By conjugating with the Fourier transform ( $\tilde{T} = \mathcal{F} \circ T \circ \mathcal{F}^{-1}$ ), they rewrite the operator using a real-valued phase $\Phi_\tau(x, \eta) = \text{Re}\,\Phi + \tau \text{Im}\,\Phi$ and a modified symbol $\sigma_\tau$ .
- Crucially, the new symbol $\sigma_\tau$ belongs to the Hörmander class $S^{m}_{1/2, 1/2}$ , which requires techniques distinct from standard pseudo-differential operator theory (necessitating Littlewood–Paley theory).
Spatial Smooth Factorization Condition (SSFC):
The analysis relies on a structural assumption on the phase function, inspired by Nicola's work on real phases.
- Definition: For a fixed $\tau$ , the real phase $\Phi_\tau$ satisfies the SSFC of rank $\kappa$ if, locally, the gradient $\nabla_x \Phi_\tau$ is constant along affine subspaces of codimension $\kappa$ .
- Geometric Implication: This implies the Hessian $d^2_x \Phi_\tau$ has rank at most $\kappa$ . The integer $\kappa$ quantifies the "flatness" or degeneracy of the phase in spatial variables.
Dyadic Decomposition and Kernel Estimates:
1. Frequency Localization: The operator is decomposed into dyadic frequency blocks using a Littlewood–Paley partition of unity.
2. Angular Decomposition: Within each dyadic shell, the phase is further decomposed into small angular caps.
3. Taylor Expansion & Integration by Parts: The phase is linearized around the center of each cap. The authors construct specific differential operators ( $L_j^v$ ) adapted to the split between non-degenerate ( $\eta'$ ) and degenerate ( $\eta''$ ) directions.
4. Kernel Bounds: By applying repeated integration by parts using these operators, they derive pointwise decay estimates for the integral kernel $K(x,y)$ . The decay is polynomial in the non-degenerate directions and controlled by the geometry of the degenerate directions.
Interpolation:
Boundedness is first established for the endpoint cases $p=1$ and $p=\infty$ using the derived kernel estimates. The result for intermediate $1 < p < \infty$ is then obtained via complex interpolation between $FL^1$ and $FL^2$ (or $FL^\infty$ ).

3. Key Contributions

Extension to Complex Phases: The paper successfully extends the boundedness theory of FIOs from real canonical relations to complex canonical relations. It demonstrates that the geometric intuition developed for real phases (specifically the role of the Hessian rank) remains valid in the complex setting.
Handling $S^{m}_{1/2, 1/2}$ Symbols: The authors provide a rigorous framework for handling the more oscillatory symbols arising from complex phases. They show that despite the symbol belonging to the less regular $S^{m}_{1/2, 1/2}$ class, the specific geometric structure (SSFC) allows for sufficient kernel decay to ensure boundedness.
Sharpness of the Order Condition: The paper proves that the derived condition on the order $m$ is sharp, meaning it cannot be improved without additional assumptions.

4. Main Results

Theorem 1.1 (Main Theorem):
Let $T$ be a Fourier integral operator of order $m$ associated with a complex canonical relation parametrized by a complex phase $\Phi$ of positive type. Assume the associated real phase $\Phi_\tau$ satisfies the Spatial Smooth Factorization Condition (SSFC) of rank $\kappa$ (where $0 \le \kappa \le d$ ).

Then, $T$ extends to a bounded operator on the Fourier Lebesgue space $FL^p(\mathbb{R}^d)$ for all $1 \le p \le \infty$ , provided that:
$m \le -\kappa \left| \frac{1}{p} - \frac{1}{2} \right|$

Key Observations from the Result:

Dependence on $\kappa$ : The required order $m$ $m$ depends linearly on the rank $\kappa$ $κ$ of the degeneracy.
- If $\kappa = d$ (maximal degeneracy), the condition is $m \le -d |1/p - 1/2|$ .
- If $\kappa = 0$ (non-degenerate, e.g., pseudo-differential operators), the condition is $m \le 0$ .
Comparison with Real Phases: For real-valued phases on $L^p$ spaces, the Seeger–Sogge–Stein theorem requires $m \le -(d-1)|1/p - 1/2|$ . The complex phase result differs significantly, particularly at the endpoints ( $p=1, \infty$ ), where the loss of derivatives is generally larger (scaling with $d$ rather than $d-1$ ) unless the phase is non-degenerate.
Sharpness: The condition is proven to be optimal; if $m$ exceeds this bound, the operator is generally unbounded.

5. Significance

Theoretical Advancement: This work bridges a gap in the microlocal analysis of partial differential equations. It provides a robust calculus for FIOs with complex phases, which are essential for studying hyperbolic equations with caustics and operators on sub-Riemannian manifolds where real phase constructions fail globally.
Global Construction: By utilizing complex phases, the authors support the construction of global parametrices that avoid the topological obstructions (caustics) inherent in real-phase theories.
Methodological Rigor: The paper refines the techniques of Nicola, Ruzhansky, and Seeger–Sogge–Stein, adapting them to the specific challenges of complex symbols and establishing a clear link between the geometric rank of the phase's Hessian and the analytic loss of derivatives in Fourier Lebesgue spaces.

In summary, the paper establishes that the boundedness of FIOs with complex phases is governed by the same geometric principles as their real counterparts, but with a modified order threshold that reflects the specific oscillatory nature of complex symbols and the rank of the spatial Hessian.