🔢 mathematics

On the Stability of Inverse Conductivity Problem for Polyhedral Inclusions under a Single Measurement

This paper establishes a logarithmic stability estimate for determining a convex polyhedral inclusion within a homogeneous isotropic medium from a single boundary measurement, utilizing singularity decomposition, propagation of smallness, and microlocal analysis.

Original authors: Chun-Hsiang Tsou

Published 2026-05-19

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Chun-Hsiang Tsou

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

Imagine you are in a dark room filled with a uniform, invisible jelly. Somewhere inside this jelly, there is a hidden object made of a different material—perhaps a piece of metal that conducts electricity much better than the jelly around it. You cannot see the object, and you cannot touch it. However, you can stand outside the room and poke the walls with a stick, measuring how the electricity flows in response.

This is the essence of the Inverse Conductivity Problem studied in this paper. The goal is to figure out the shape and location of that hidden object just by looking at the electrical signals on the outside wall.

The Specific Challenge: The "Polyhedral" Mystery

Most previous studies assumed the hidden object was smooth, like a ball or an egg. But in the real world, objects often have sharp corners and flat faces, like a die, a pyramid, or a chunk of rock. These are called polyhedral inclusions.

The author, Chun-Hsiang Tsou, tackles a very difficult version of this problem:

The Shape: The hidden object is a convex polyhedron (a shape with flat faces and sharp corners, like a soccer ball made of flat panels).
The Data: You only get one single measurement. Usually, scientists might take hundreds of measurements from different angles to build a picture. Here, you only get one snapshot of the electrical flow from one side of the room.
The Goal: Prove that even with this limited data, you can mathematically guarantee how close your guess of the object's shape is to the real thing.

The Main Result: A "Logarithmic" Promise

The paper proves a stability estimate. In plain English, this means: "If your measurement has a tiny bit of error, your guess of the object's shape won't be wildly wrong; it will only be slightly off."

However, the relationship isn't perfect. The paper shows that the error in your guess grows logarithmically with the measurement error.

Analogy: Imagine trying to tune a radio. If the signal is slightly fuzzy (small error), you might have to turn the dial a little bit to find the station. But if the signal is very fuzzy, you might have to turn the dial a huge amount to find the right spot.
The Catch: Because the relationship is logarithmic, the problem is "ill-posed." This means that while a solution exists and is unique, it is extremely sensitive. A tiny mistake in your measurement can lead to a larger (though still controlled) mistake in your reconstruction. The paper provides a mathematical formula that puts a "ceiling" on how bad this mistake can get.

How Did They Solve It? The Three Magic Tools

To prove this, the author didn't just guess; they used three sophisticated mathematical "tools" to dissect the problem:

1. Singularity Decomposition (The "Corner Detector")

When electricity flows around a smooth ball, it behaves nicely. But when it hits a sharp corner or an edge of a polyhedron, it gets "jumpy" or "singular."

The Metaphor: Think of water flowing around a smooth rock versus a jagged cliff. The water swirls violently near the jagged edges.
The Tool: The author breaks the mathematical solution into two parts: a "smooth" part that behaves normally, and a "singular" part that describes exactly how the electricity behaves near those sharp corners and edges. By understanding the specific "signature" of these sharp corners, they can use them as landmarks to locate the object.

2. Propagation of Smallness (The "Whisper Chain")

This tool answers the question: "If I know the signal is weak on the outside wall, how weak is it deep inside the room?"

The Metaphor: Imagine a whisper in a large hall. If you hear a very quiet whisper at the door, you know the person speaking must be very quiet or very far away. This tool allows the author to trace the "whisper" of the measurement error from the outside wall all the way to the hidden object, proving that the error doesn't magically disappear or explode as it travels inward.

3. Microlocal Analysis (The "Flashlight")

This is a way of zooming in on specific parts of the object (like a single edge or vertex) to see how the signal behaves there.

The Metaphor: Instead of looking at the whole dark room, the author shines a specialized, tunable flashlight on just one edge of the hidden object. They use a special "test function" (a mathematical flashlight beam) that is designed to react strongly to that specific edge. By analyzing how the beam interacts with the edge, they can measure the distance between the real object and their guess.

The Conclusion

The paper concludes that even with just one measurement, it is mathematically possible to determine the shape of a hidden, sharp-cornered object with a guaranteed level of accuracy.

While the accuracy isn't perfect (due to the logarithmic nature of the error), the paper provides a rigorous "safety net." It tells us exactly how much our reconstruction might be off based on the noise in our data. This is a significant step forward because it moves beyond smooth, idealized shapes to handle the sharp, complex geometries found in real-world materials, all while relying on minimal data.

In short: The author proved that you can find a hidden, jagged rock inside a block of jelly by poking the jelly once, provided you have a very sophisticated mathematical map of how electricity behaves around sharp corners.

Technical Summary: Stability of Inverse Conductivity Problem for Polyhedral Inclusions Under a Single Measurement

Problem Formulation
This paper addresses the inverse conductivity problem, a specific case of the classical Calderón problem, within a bounded Lipschitz domain $\Omega \subset \mathbb{R}^3$ . The objective is to determine the shape and location of an unknown inclusion $D$ embedded in a homogeneous isotropic medium. The conductivity distribution $\sigma$ is modeled as a piecewise constant function with a jump across the interface $\partial D$ :
$\sigma = 1 + (k-1)\chi_D$
where $k > 1$ represents the conductivity contrast and $\chi_D$ is the indicator function of the inclusion. The inclusion $D$ is assumed to be a convex polyhedron. The inverse problem is formulated using a single partial boundary measurement: the Neumann data $\partial_\nu u|_{\Gamma_0}$ on an open subset $\Gamma_0 \subset \partial\Omega$ , given a non-trivial Dirichlet boundary condition $f \in H^{1/2}(\partial\Omega)$ . The forward problem is governed by the elliptic equation $\text{div}(\sigma \nabla u) = 0$ .

Methodology and Technical Ingredients
The analysis relies on three primary mathematical tools to bridge the gap between the boundary measurement error and the geometric reconstruction error:

Singularity Decomposition: The authors utilize the theory of elliptic equations in non-smooth domains to decompose the solution $u$ near the polyhedral interface into a regular part and singular parts associated with vertices and edges.
- Near a vertex $v$ , the solution behaves as a linear combination of singular functions of the form $\rho^\lambda \Phi(\vartheta)$ , where $\lambda$ are singular exponents determined by the local geometry and conductivity contrast.
- Near an edge $E$ , the solution involves terms like $r^\mu \phi(\theta)$ , where $\mu$ are edge singular exponents.
- The paper imposes generic assumptions to ensure the leading singular coefficients are non-degenerate and that logarithmic terms (arising from integer exponents) are absent.
Propagation of Smallness: This technique is used to estimate the difference between solutions corresponding to two different inclusions ( $u$ and $u'$ ) in the interior of the domain based on the discrepancy in their Cauchy data on the accessible boundary $\Gamma_0$ . By iteratively applying the three-sphere inequality along a chain of balls connecting the boundary to the interior, the authors derive logarithmic stability estimates for the solution difference $u - u'$ and its gradient in regions away from the inclusion.
Microlocal Analysis and Complex Geometrical Optics (CGO): To derive the final stability estimate, the authors construct a specific integral identity involving a CGO solution $u_0(x) = e^{\zeta \cdot (x-x_O)}$ with a complex vector $\zeta$ . This test function is inserted into a variational formulation over a carefully selected local cylindrical region adjacent to an edge of the polyhedron. The analysis involves estimating the integral identity by separating contributions from vertex singularities, edge singularities, and the regular part, utilizing the specific decay properties of the CGO solution.

Key Results
The principal result of the paper is a quantitative logarithmic stability estimate for the Hausdorff distance between two convex polyhedral inclusions $D$ and $D'$ . Let $\varepsilon$ denote the measurement error in the $H^{-1/2}(\Gamma_0)$ norm of the difference in Neumann data. Under specific geometric and analytic assumptions (including a priori bounds on the polyhedron's geometry, the distance of the inclusion from the boundary, and the conductivity contrast), the authors prove:
$d_H(D, D') \leq C (\ln |\ln \varepsilon|)^{-\kappa}$
where $C$ and $\kappa$ are positive constants depending only on the a priori data (such as the domain geometry, the minimum edge length, vertex/edge opening angles, and the conductivity contrast $k$ ).

The proof proceeds by:

Establishing that the Hausdorff distance is attained at a vertex.
Using the integral identity to relate the boundary measurement error to the local behavior of the solution near the edge.
Balancing the decay of the CGO solution against the growth of the singular terms to optimize the parameter $\tau$ in the test function.
Combining the propagation of smallness estimates with the local integral estimates to solve for the geometric distance $h$ (related to $d_H$ ) in terms of $\varepsilon$ .

Significance and Contribution
The paper contributes to the literature on inverse problems by addressing the stability of determining polyhedral inclusions from a single boundary measurement. While uniqueness results for polyhedral inclusions in three dimensions have been established (e.g., in [4]), stability results for this specific configuration under a single measurement were previously open.

The authors position their work within the context of the Calderón problem, noting that while smooth conductivities can be stably recovered with infinitely many measurements (Dirichlet-to-Neumann map), piecewise constant conductivities with jumps are more realistic for applications like electrical impedance tomography (EIT). Previous stability results for inclusions often relied on multiple measurements or local stability assumptions (e.g., $C^1$ perturbations). This work demonstrates that even with a single measurement, a logarithmic stability estimate can be achieved for polyhedral shapes by exploiting their specific geometric singularities. The methodology combines singularity analysis, unique continuation, and microlocal techniques to handle the non-smooth nature of the polyhedral interface, providing a rigorous theoretical foundation for the reconstruction of such inclusions.