🔢 mathematics

The weak (1,1) boundedness of Fourier integral operators with complex phases

This paper establishes the weak (1,1) boundedness of Fourier integral operators of order $-(n-1)/2$ associated with canonical relations parametrized by complex phase functions, a result that cannot be derived from the existing theory for real-valued phases.

Original authors: Duván Cardona, Michael Ruzhansky

Published 2026-02-18

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Duván Cardona, Michael Ruzhansky

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: Predicting the Weather with a Broken Thermometer

Imagine you are trying to predict the weather. You have a complex machine (a Fourier Integral Operator) that takes in data about the wind and temperature and tries to tell you what the weather will be like tomorrow.

In the world of mathematics, this machine processes "waves" of information. For a long time, mathematicians knew how to handle these machines when the data was "real" and straightforward (like a standard thermometer reading). They knew that if you put a messy, chaotic input into the machine, the output would stay reasonably controlled. This is called boundedness.

However, in the real world (and in advanced physics), things aren't always "real." Sometimes the data is "complex," meaning it has a hidden, imaginary component (like a thermometer that also measures a secret, invisible energy). This paper tackles a specific, difficult question: If we use this machine with this "complex" hidden energy, does it still keep the output under control?

The authors, Duván Cardona and Michael Ruzhansky, say: "Yes, it does." They proved that even with this complex, hidden energy, the machine won't explode or produce infinite chaos.

The Cast of Characters

To understand the proof, let's meet the main players using metaphors:

The Operator (The Machine): This is the tool that transforms input data into output. Think of it as a giant, high-tech blender.
The Phase (The Recipe): Every time the blender spins, it follows a specific recipe (a mathematical function called a "phase").
- Real Phase: A standard, predictable recipe.
- Complex Phase: A recipe that includes a secret ingredient (the "imaginary" part) that changes how the ingredients mix. This makes the math much harder because the secret ingredient can make things dampen or explode in weird ways.
The "Weak (1,1)" Rule: This is the safety standard.
- Imagine you have a bucket of sand (the input). If you pour it through the blender, you don't want the output to be a mountain of sand that spills everywhere (infinite chaos).
- The "Weak (1,1)" rule is a promise: "Even if the input is a bit messy, the output won't be too messy. It might spill a little, but it won't flood the whole house."

The Problem: Why Was This Hard?

For decades, mathematicians knew how to prove this safety rule for the "Real Phase" machines (thanks to a brilliant proof by Terence Tao in 2004).

But when they tried to apply the same logic to the "Complex Phase" machines, it failed.

The Analogy: Imagine you have a recipe for a cake that works perfectly with real eggs. You try to use the same recipe with "imaginary eggs" (a metaphor for the complex phase). The cake doesn't just taste different; the whole structure collapses.
The "Complex Phase" introduces a new kind of friction and decay that the old math tools couldn't handle. You can't just pretend the secret ingredient doesn't exist; you have to account for it.

The Solution: A New Strategy

The authors didn't try to force the old recipe to work. Instead, they invented a new strategy by combining two different approaches:

1. The "Split and Conquer" Strategy (Tao's Method)

Terence Tao previously solved the "Real Phase" problem by splitting the machine into two parts:

The "Degenerate" Part: The part of the machine where the recipe is "flat" or boring. It doesn't do much fancy work.
The "Non-Degenerate" Part: The part where the recipe is "curvy" and active. This is where the magic (and the danger) happens.

Tao proved that the "Boring" part is safe, and the "Active" part can be broken down into a safe filter and a safe average.

2. The "Secret Ingredient" Adjustment (The Authors' Twist)

Cardona and Ruzhansky realized they could use Tao's split strategy, but they had to be very careful with the "Complex Phase."

The Challenge: In the complex world, the "Active" part of the machine has a weird, oscillating secret ingredient ( $e^{-(1+i\tau)\text{Im}(\Phi)}$ ). If they tried to treat this like a normal ingredient, the math would break.
The Fix: They decided to hide the secret ingredient inside the "Average" part of the machine.
- Imagine the machine is a factory. Instead of trying to fix the weird ingredient in the main assembly line, they moved it to the "Quality Control" station (the averaging operator).
- They proved that even with this weird ingredient in Quality Control, the station could still handle the workload without overflowing.

The "Ellipsoid" Analogy

One of the most clever parts of their proof involves how they look at the "curvature" of the recipe.

The Old Way: Imagine trying to fit a square peg into a round hole. You look at the data in a grid (like a checkerboard).
The New Way: The authors realized that because of the "Complex Phase," the data doesn't fit in a grid. It fits better in ellipses (stretched circles).
They created a new way to slice the data into these "elliptical slices." This allowed them to see that the "flat" parts of the machine were actually very safe, and the "active" parts could be tamed.

Why Does This Matter? (The "So What?")

You might ask, "Who cares about imaginary eggs in a blender?"

This math is the foundation for understanding waves in the real world.

Quantum Mechanics: Particles behave like waves with complex phases.
Seismology: Earthquakes create waves that travel through the earth.
Medical Imaging: MRI machines use these principles to create pictures of the inside of your body.

By proving that these "Complex Phase" machines are safe (bounded), the authors ensure that the mathematical models we use to predict earthquakes, design lasers, or understand quantum particles are reliable. They proved that even when the math gets "complex" and weird, the universe (and our models of it) remains stable and predictable.

Summary

The Goal: Prove that a specific type of mathematical machine stays under control even when it uses "complex" (imaginary) ingredients.
The Obstacle: Old methods failed because the "imaginary" ingredients changed the rules of the game.
The Breakthrough: The authors took a proven strategy for "real" machines, modified it to hide the "imaginary" ingredients in a safe zone, and used a new way of slicing the data (ellipses instead of grids) to prove everything stays stable.
The Result: We now know that these complex mathematical tools are safe to use for modeling real-world phenomena like waves and vibrations.

1. Problem Statement

The paper addresses the boundedness properties of Fourier Integral Operators (FIOs) associated with complex-valued phase functions. Specifically, the authors aim to prove the weak $(1,1)$ boundedness for FIOs of order $m = -(n-1)/2$ .

Context: In the theory of FIOs with real-valued phases, it is a fundamental result (Seeger, Sogge, and Stein, 1991) that operators of order $-(n-1)/2$ map the Hardy space $H^1$ to $L^1$ . Subsequently, Terence Tao (2004) proved that these same operators satisfy the weak $(1,1)$ inequality (boundedness from $L^1$ to $L^{1,\infty}$ ).
The Gap: While the $L^p$ ( $1 < p < \infty$ ) and $H^1 \to L^1$ boundedness for FIOs with complex phases were established by the second author (Ruzhansky) in previous work, the weak $(1,1)$ endpoint remained open.
The Challenge: The result for real phases cannot be directly derived for complex phases. When attempting to reduce a complex phase operator to a real one (via a transformation $\Phi_\tau = \text{Re}(\Phi) + \tau \text{Im}(\Phi)$ ), the resulting symbol falls into the Hörmander class $S^0_{1/2, 1/2}$ . Standard pseudo-differential operator theory suggests that operators in this class with order $m$ are only weak $(1,1)$ bounded if $m \leq -n/4$ . However, the required order for FIOs is $-(n-1)/2$ , which is strictly better (more negative) than $-n/4$ only for $n > 1$ , but the reduction introduces a loss of regularity that standard methods cannot handle without specific modifications.

2. Methodology

The authors employ a hybrid strategy combining two distinct approaches:

Melin-Sjögstrand Reduction (Complex to Real):
They utilize the technique developed by Melin and Sjögstrand to rewrite the complex phase operator $T$ in terms of a real phase function $\Phi_\tau(x, \xi) = \text{Re}(\Phi) + \tau \text{Im}(\Phi)$ and a modified symbol. This introduces an oscillating damping term $e^{-(1+i\tau)\text{Im}(\Phi)}$ .
$Tf(x) = \int e^{2\pi i \Phi_\tau(x, \xi)} a(x, \xi) e^{-(1+i\tau)\text{Im}(\Phi)} \hat{f}(\xi) d\xi$
Tao's Factorization Strategy (Real Phase Adaptation):
They adapt Terence Tao's proof for the real-valued case, which involves decomposing the operator into degenerate and non-degenerate components based on the curvature of the phase function.
- Decomposition: The operator $T$ $T$ is split into $T = T_{\text{deg}} + T_{\text{non-deg}}$ $T = T_{deg} + T_{non-deg}$ .
  - $T_{\text{deg}}$ : Corresponds to regions where the Hessian of the phase (curvature) is small (near-degenerate).
  - $T_{\text{non-deg}}$ : Corresponds to regions where the curvature is significant.
- Factorization of Non-Degenerate Part: The non-degenerate part is factorized as $T_{\text{non-deg}} = S \circ A + E$ $T_{non-deg} = S \circ A + E$ , where:
  - $S$ is a pseudo-differential operator of order 0.
  - $A$ is an averaging operator.
  - $E$ is an error operator.

Key Technical Modifications for Complex Phases:

Handling the Symbol Class: In the real case, the symbol of $S$ is in $S^0_{1,0}$ . In the complex case, the natural reduction places the symbol in $S^0_{1/2, 1/2}$ . To avoid the regularity loss associated with this class (which would require order $\leq -n/4$ ), the authors keep the original symbol $a$ within the pseudo-differential part $S$ and absorb the oscillating term $e^{-(1+i\tau)\text{Im}(\Phi)}$ into the averaging operator $A$ .
Continuous Dyadic Decomposition: Instead of Tao's discrete "second dyadic decomposition" (using disks), the authors introduce a continuous partition of unity using a family of ellipsoids. These ellipsoids adapt to the local spectrum of the Hessian of the phase function. This allows them to avoid Kakeya-type covering lemmas and control the eccentricity and orientation of the regions where the phase degenerates.
Strictly Positive Type: The proof is first reduced to the case where the phase is of "strictly positive type" (i.e., $\text{Im}(\Phi) > \kappa_0 > 0$ ), ensuring exponential decay in the kernel estimates.

3. Key Contributions

Proof of Weak $(1,1)$ for Complex Phases: The paper establishes that FIOs of order $-(n-1)/2$ with complex phases of positive type are of weak $(1,1)$ type.
Adaptation of Tao's Method: It successfully extends Tao's factorization technique to the complex setting, overcoming the regularity issues inherent in the $S^0_{1/2, 1/2}$ symbol class by redefining the factorization components ( $S$ and $A$ ).
Geometric Analysis of Degeneracy: The introduction of a continuous decomposition based on ellipsoids adapted to the Hessian of the complex phase provides a robust framework for analyzing the degenerate components in the presence of complex damping.
Error Control: The authors provide a rigorous estimate for the error operator $E$ , showing it is bounded on $L^1$ despite the presence of the complex oscillating term, by utilizing stationary phase arguments and careful symbol estimates.

4. Main Results

Theorem 1.7 (Main Result):
Let $T$ be a Fourier integral operator of order $\mu = -(n-1)/2$ associated with a canonical relation parametrized by a complex phase $\Phi$ of positive type (specifically $\text{Im}[\Phi(x, \xi)] > 0$ when $|\xi| \neq 0$ ) satisfying the local graph condition. Then, $T$ is of weak $(1,1)$ type.

Corollary 5.1 (Application to PDEs):
The result is applied to the Cauchy problem for first-order classical pseudo-differential operators with complex characteristics (non-negative imaginary part of the principal symbol).

If the initial data $f$ belongs to the Sobolev space $L^1_{s + (n-1)/2}(X)$ , the solution $v(t, \cdot)$ at any fixed time $t$ belongs to the local weak Sobolev space $L^{1,\infty}_{s, \text{loc}}(X)$ .
This provides a priori estimates for hyperbolic equations with complex characteristics.

5. Significance

Completeness of Theory: This work completes the endpoint boundedness theory for FIOs with complex phases, mirroring the established theory for real phases. It confirms that the critical order $-(n-1)/2$ is sharp for weak $(1,1)$ boundedness even in the complex setting.
Methodological Advancement: The paper demonstrates that the factorization approach of Tao is robust enough to handle the additional complexities of complex phases (damping terms and different symbol classes) through strategic redefinition of the operators involved.
Applications: The results have direct implications for the regularity theory of solutions to partial differential equations with complex characteristics, such as those arising in the study of the oblique derivative problem or operators with non-real principal symbols. It bridges the gap between the geometric analysis of complex canonical relations and the functional analytic properties of the associated operators.

In summary, Cardona and Ruzhansky successfully extend a landmark result in harmonic analysis (Tao's weak $(1,1)$ bound for FIOs) to the complex domain, utilizing a sophisticated blend of symplectic geometry, oscillatory integral asymptotics, and refined microlocal analysis.