Scattering rigidity for Hamiltonian systems with an application to Finsler geometry

Imagine you are standing outside a mysterious, foggy room. You cannot see inside, but you can throw balls (or send light beams) into the room and watch how they bounce off the walls and come back out. By measuring where they come out, how fast they return, and what angle they hit the wall, you want to figure out exactly what the inside of the room looks like.

This is the core idea of the paper "Scattering Rigidity for Hamiltonian Systems." The authors, Nikolas Eptaminitakis and Plamen Stefanov, are solving a complex mathematical puzzle about how much we can learn about the "shape" of a space just by watching how things move through it.

Here is a breakdown of their work using simple analogies.

1. The Setup: The "Black Box" and the "Map"

Think of the "room" as a mathematical space called a manifold (which could be a curved surface, a 3D space, or something even stranger). Inside this space, there are invisible rules that dictate how things move. In physics, these rules are often described by something called a Hamiltonian.

The Analogy: Imagine the Hamiltonian is the "law of the land" for a specific type of vehicle. In a normal city (Riemannian geometry), cars drive on flat roads, and the distance between two points is straightforward. But in this paper, the "roads" might be warped, or the cars might have different speeds depending on which direction they face (this is called Finsler geometry).
The Goal: The authors want to know: If I give you a list of every possible trip a car could take from the entrance to the exit (the scattering relation) and how long each trip took (the travel times), can you reconstruct the exact "law of the land" (the Hamiltonian) that governs the movement?

2. The Two Scenarios: High Energy vs. Zero Energy

The paper splits the problem into two distinct "worlds," depending on the energy of the moving object.

World A: The "High Energy" World (Positive Energy)

Imagine throwing a ball with plenty of force. It zips through the room, hits the walls, and bounces out.

The Discovery: The authors prove that if two different "laws of the land" produce the exact same entry/exit points and travel times, then those two laws are actually the same, just viewed through a different "lens."
The "Lens" (Canonical Transformation): Think of this like looking at a sculpture through a funhouse mirror. The sculpture looks distorted, but if you know the rules of the mirror, you can figure out the original shape. The authors show that the only way two different systems can look identical from the outside is if they are related by a specific kind of mathematical "mirror" that doesn't change the boundary (the doorways).
The Catch: In the real world (like Riemannian geometry), we usually want to know if the shape of the room itself is unique. This paper shows that in the broader mathematical world, the "shape" might be unique only up to these fancy mirrors.

World B: The "Zero Energy" World (Light Rays)

Now, imagine a beam of light. Light doesn't have "mass" in the traditional sense, and in this mathematical model, it travels on a special path called a "null bicharacteristic."

The Problem: You can't really measure "time" for light in the same way you do for a ball, because light travels at the speed limit of the universe. Instead of travel time, the authors use a different measuring stick: a defining function.
The Analogy: Imagine you are trying to map a dark cave using only the echoes of your voice. You can't time the echo perfectly, but you know exactly where the echo comes from. The authors show that even without precise timing, if you know the pattern of where the light enters and exits, you can still reconstruct the "light cone" (the shape of the paths light can take).
The Result: They prove that the "light map" is unique up to a scaling factor. It's like saying, "I can tell you the shape of the cave, but I can't tell you if the cave is 10 feet wide or 100 feet wide without a ruler." However, the geometry of the paths is fixed.

3. The "X-Ray" Trick

To solve these puzzles, the authors use a technique called Linearization.

The Analogy: Imagine you are trying to figure out the shape of a bumpy hill. Instead of trying to map the whole bumpy hill at once, you imagine the hill is almost flat, and you only look at the tiny bumps. You measure how those tiny bumps change the path of a rolling ball.
The Math: They turn the complex, non-linear problem into a simpler "X-ray transform." Imagine shining an X-ray through the room. The X-ray beam passes through the object, and the amount it gets "absorbed" tells you about the density of the object. The authors prove that they can "invert" this X-ray transform—meaning, if they know the total absorption along every possible path, they can mathematically reconstruct the density of the object inside.

4. The Real-World Application: Finsler Geometry and Elasticity

Why does this matter? The authors apply this to Finsler manifolds.

The Real World: Think of a piece of wood or a crystal. If you send a sound wave through it, the wave travels faster in some directions than others. This is called anisotropy.
The Connection: In a normal rubber ball (isotropic), sound travels the same speed in all directions. In a crystal, the "speed limit" depends on the direction. This is exactly what a Finsler metric describes.
The Breakthrough: The authors show that if you can measure how seismic waves (or sound waves) enter and exit a piece of material, you can figure out the internal structure of that material. This is crucial for anisotropic elasticity—helping engineers understand how materials like composite fibers or crystals behave under stress without having to cut them open.

Summary: The Big Picture

The paper is a masterclass in inverse problems.

The Question: Can we see the invisible? (Can we deduce the internal rules of a system just by watching how things enter and exit?)
The Answer: Yes, mostly.
- If things have "energy" (like balls), the internal rules are unique, provided we account for certain mathematical "mirrors."
- If things are "light-like" (zero energy), the paths are unique, though the scale might be ambiguous.
The Tool: They developed a new mathematical "X-ray" that works on these complex, curved spaces, allowing them to reverse-engineer the geometry of the space from the data of the paths.

In short, the authors have built a mathematical "CT scanner" for the abstract shapes of the universe, proving that if you listen carefully to how things bounce around, you can reconstruct the entire room.

Here is a detailed technical summary of the paper "Scattering Rigidity for Hamiltonian Systems with an Application to Finsler Geometry" by Nikolas Eptaminitakis and Plamen Stefanov.

1. Problem Statement

The paper investigates scattering rigidity and boundary rigidity (lens rigidity) problems for Hamiltonian systems defined on the punctured cotangent bundle $T^*M \setminus 0$ of a compact manifold $M$ with boundary. The central question is: To what extent does the scattering relation (the map between incoming and outgoing boundary covectors) and the travel times between boundary points uniquely determine the underlying Hamiltonian $H(x, \xi)$ ?

The authors address two distinct regimes:

Positive Energy Levels ( $H = E > 0$ ): Relevant to Riemannian and pseudo-Riemannian metrics, and anisotropic elasticity.
Zero Energy Level ( $H = 0$ ): Relevant to operators of real principal type, such as the wave equation in Lorentzian geometry (light rays) and propagation of singularities for pseudo-differential operators.

The problem is formally underdetermined because the data (scattering relation) depends on $2n-2 $variables, while the Hamiltonian depends on$ 2n-1$ variables. The authors aim to characterize the non-uniqueness (gauge freedom) and prove rigidity up to this gauge.

2. Methodology

The authors employ a phase space approach combining symplectic geometry, microlocal analysis, and inverse problem techniques.

A. Positive Energy Levels ( $H > 0$ )

Scattering Relation & Travel Times: They define the scattering relation $S$ mapping incoming boundary covectors to outgoing ones, and travel times $\ell$ .
Canonical Transformations: The core methodology relies on the fact that if two Hamiltonians share the same scattering data, they are related by a homogeneous canonical transformation $\kappa$ (a symplectic diffeomorphism) that preserves the boundary structure.
Linearization: They linearize the travel times $T(x, y)$ with respect to the Hamiltonian. Unlike the Riemannian case where geodesics are critical points of the energy functional, the authors prove that Hamiltonian curves are not critical points of the Hamiltonian energy functional in general. However, the linearization still yields an X-ray transform ( $Xf$ ) along Hamiltonian curves.
Inversion: They invert the X-ray transform modulo a gauge (functions of the form $X_H \phi$ where $\phi$ vanishes on the boundary).

B. Zero Energy Level ( $H = 0$ )

Conic Structure: At zero energy, the level set $H^{-1}(0)$ is conic and codimension-one. Travel times cannot be naturally defined without a time function.
Defining Functions: Instead of travel times, the authors use defining functions $r(x, y)$ for the set of boundary pairs connected by null bicharacteristics.
Light Ray Transform: The linearization of these defining functions leads to the "Hamiltonian light ray transform" ( $L$ ), an X-ray transform over null rays.
Gauge Freedom: They characterize the kernel of $L$ and show that the scattering relation determines the Hamiltonian flow up to a reparameterization (multiplication by a positive function $\mu$ ) and a canonical transformation.

C. Application to Finsler Geometry

Legendre Transform: They utilize the Legendre transform to map Finsler metrics (on the tangent bundle $TM$ ) to strictly convex, homogeneous Hamiltonians (on the cotangent bundle $T^*M$ ).
Reduction: This allows them to apply their Hamiltonian rigidity results to Finsler manifolds.

3. Key Contributions and Results

A. Semi-Global Rigidity for Positive Energy (Theorem 2.1)

Result: If two Hamiltonians $H$ $H$ and $\tilde{H}$ $\tilde{H}$ have the same scattering relation $S$ $S$ and travel times $\ell$ $ℓ$ on a neighborhood of the boundary, there exists a homogeneous canonical transformation $\kappa$ $κ$ such that:
1. $\kappa$ maps the domain of one system to the other.
2. $\kappa$ is the identity on the boundary (in the sense of preserving boundary covectors).
3. $H = \tilde{H} \circ \kappa^{-1}$ .
Significance: This provides a complete characterization of the non-uniqueness. The data determines the Hamiltonian up to a pullback by a canonical transformation fixing the boundary. This is a stronger result than just "up to diffeomorphism," as $\kappa$ acts on the phase space and may not be the lift of a base diffeomorphism.

B. Linearization and X-ray Transform (Section 3)

Variational Formula: The authors derive a new variational formula (Lemma 3.1) showing that the first variation of the travel time involves an integral of the perturbation of the Hamiltonian along the curve.
Non-Criticality: They explicitly demonstrate that Hamiltonian curves are not critical points of the energy functional in the general Hamiltonian setting (unlike geodesics in Riemannian geometry).
Inversion: They prove that the linearized travel time map is the X-ray transform $X$ , and its kernel consists of functions of the form $X_H \phi$ where $\phi$ vanishes on the boundary.

C. Zero Energy Rigidity (Theorem 4.1 & 4.2)

Result: For operators of real principal type, the scattering relation determines the zero energy hypersurface and the flow up to a conformal factor $\mu$ and a canonical transformation.
Light Ray Transform: They define the Hamiltonian light ray transform $L$ and prove its injectivity modulo the gauge $X_H \phi$ .
Generating Functions: They show that any defining function of the set of connected boundary pairs serves as a generating function for the scattering relation, and its linearization yields the light ray transform.

D. Finsler Rigidity (Theorem 5.1)

Result: The authors prove semi-global lens rigidity for non-trapping Finsler manifolds.
Gauge Group: The group of transformations preserving the boundary data consists of canonical transformations composed with Legendre transforms.
Crucial Distinction: They highlight that the gauge group for Finsler rigidity is larger than the group of base diffeomorphisms. There exist canonical transformations that preserve the scattering data but do not arise from diffeomorphisms of the base manifold. This implies that Finsler rigidity cannot be solved simply by viewing it as a problem of recovering a metric up to base diffeomorphism.

4. Significance and Implications

Unification of Inverse Problems: The paper unifies the study of rigidity for Riemannian metrics, Lorentzian metrics (light rays), and general Hamiltonian systems under a single phase-space framework.
Clarification of Gauge Freedom: It precisely identifies the "gauge" in inverse problems. For positive energy, it is a canonical transformation fixing the boundary. For zero energy, it includes a conformal factor (reparameterization of time).
Finsler Geometry: The application to Finsler geometry is a major contribution. It resolves the rigidity problem for a broad class of Finsler metrics and clarifies that the non-uniqueness is intrinsic to the phase space structure, not just the base geometry. This has direct implications for anisotropic elasticity, where wave speeds depend on direction in a non-Riemannian way.
Microlocal Connection: The Appendix connects these geometric results to the Dirichlet-to-Neumann (DN) map and the propagation of singularities for hyperbolic PDEs, showing that the scattering relation is the canonical relation of the DN map.
Methodological Innovation: The use of generating functions for the scattering relation and the specific variational formulas for travel times provide new tools for analyzing linearized inverse problems in non-Riemannian settings.

In summary, the paper establishes that scattering data determines a Hamiltonian system uniquely up to a specific class of symplectic symmetries (canonical transformations) that fix the boundary, providing a comprehensive solution to the rigidity problem for both positive and zero energy levels, with significant applications to Finsler geometry and anisotropic wave propagation.