Abstract
We prove that the reconstruction of a certain type of length spaces from their travel time data on a closed subset is Lipschitz stable. The travel time data is the set of distance functions from the entire space, measured on the chosen closed subset. The case of a Riemannian manifold with boundary with the boundary as the measurement set appears is a classical geometric inverse problem arising from Gel'fand's inverse boundary spectral problem. Examples of spaces satisfying our assumptions include some non-simple Riemannian manifolds, Euclidean domains with non-trivial topology, and metric trees.
Introduction
A fundamental challenge in geometric inverse problems is determining the internal structure of an object based solely on measurements taken at its boundary. This problem, often framed as reconstructing a manifold from "travel time data," has significant implications for fields like seismic imaging and medical tomography. While previous work established that such data can uniquely identify certain smooth manifolds, a critical gap remained regarding the stability of these reconstructions. In practical applications, measurements are inherently noisy; a reconstruction method is only viable if small errors in data do not lead to disproportionately large errors in the resulting model.
Methodology and Theoretical Framework
This research extends the study of travel time data from smooth Riemannian manifolds to the broader category of length spaces. To establish Lipschitz stability, the authors introduce several key theoretical constructs:
Local Isometric Embeddings (FLIE and BLIE)
The paper defines "Forward/Backward Locally Isometrically Embeddable" (FLIE/BLIE) conditions. These properties ensure that the travel time map—which maps points in the space to their respective distance functions—behaves as a local isometry. This local behavior is shown to be sufficient for global stability.
Quantifying Similarity
To measure the distance between reconstructed spaces, the authors utilize the Gromov–Hausdorff distance. Furthermore, they introduce the $ε$-truncated Gromov–Hausdorff distance, which allows for comparing spaces at a specific scale $ε$. This innovation is crucial for proving stability in scenarios where global diameter bounds are unavailable or inapplicable.
Key Results
The core contribution of this work is the rigorous proof of Lipschitz stability across various geometric configurations.
- Quantitative Stability (Theorem 6): For spaces with a diameter bounded by $D$, the Gromov–Hausdorff distance between two spaces is bounded by a constant multiple $(\frac{2D}{\epsilon} + 1)$ of the Hausdorff distance between their respective travel time datasets.
- Scale-Independent Stability (Theorem 9): By employing the $ε$-truncated distance, the authors demonstrate that stability can be maintained even without a global diameter bound, making the results applicable to a wider variety of geometries.
- Smooth Reconstruction: For Finsler or Riemannian manifolds, the research proves that travel time data determines the manifold up to a smooth isometry, provided the local isometry condition holds.
Scope and Applications
Unlike previous studies that often required "simple" manifolds (those without trapped paths or internal focal points), this paper demonstrates that stability holds for "non-simple" cases, including:
- Riemannian and Finsler manifolds with non-convex boundaries.
- Metric trees and graphs.
- Euclidean domains with non-trivial topology (e.g., an annulus).
- Busemann G-spaces.
Conclusion and Future Directions
By proving that the relationship between travel time data and the underlying space is Lipschitz continuous, this paper provides a robust theoretical foundation for geometric reconstruction. However, challenges remain. Verifying the BLIE condition for complex, high-codimension measurement sets (such as sparse sensor arrays) is difficult. Future work will focus on developing numerical algorithms that can translate these theoretical Lipschitz bounds into practical, robust 3D reconstruction tools.
Summary updated