🔢 mathematics

Stable determination of the first order perturbation of the biharmonic operator from partial data

This paper establishes logarithmic stability estimates for the first-order perturbation of the biharmonic operator in dimensions three and higher, using a partial Dirichlet-to-Neumann map under the assumption that lower-order perturbations are known near the boundary.

Original authors: Boya Liu, Salem Selim

Published 2026-06-18

📖 5 min read🧠 Deep dive

CC0 1.0

Original authors: Boya Liu, Salem Selim

Original paper dedicated to the public domain under CC0 1.0 (http://creativecommons.org/publicdomain/zero/1.0/). ✨ 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 have a mysterious, solid block of material (like a block of ice or a piece of wood) that you cannot cut open. You want to know exactly what is happening inside it—specifically, you want to map out two invisible forces: a "wind" blowing through the material (the vector field A) and a "temperature" or "pressure" field (the function q).

The only way to learn about the inside is by poking the surface. You can push on the surface (apply a Dirichlet condition) and measure how the surface reacts or "ripples" (the Neumann condition). However, there's a catch: you can only touch and measure a tiny, arbitrary patch of the surface, not the whole thing. Furthermore, you already know the properties of the material right near the surface, but the mystery lies deep in the center.

This paper is about a mathematical detective story: Can we figure out what's happening deep inside the block just by poking a tiny spot on the outside?

The Mathematical "Machine"

The authors are studying a specific mathematical machine called the Biharmonic Operator.

The Analogy: Think of a thin, elastic metal plate (like a drum skin, but stiffer). If you press on it, it bends. The "Biharmonic" math describes how this plate settles into a stable shape.
The Perturbations: The paper looks at two invisible things messing with this plate:
1. The First-Order Perturbation (A): Imagine a wind blowing across the plate, pushing it sideways.
2. The Zeroth-Order Perturbation (q): Imagine a hidden weight or pressure pushing the plate up or down.

The goal is to reverse-engineer these invisible forces (A and q) based only on the measurements taken from a small part of the edge.

The Big Challenge: "Partial Data"

Usually, to solve these puzzles, you need to measure the entire surface. But in the real world, you often can't. Maybe the object is buried, or part of it is blocked. This paper tackles the "Partial Data" problem: What if we only have a tiny window into the object?

The Main Discoveries (The "Stability" Results)

The authors prove that you can solve this puzzle, but with a warning: The solution is unstable.

In math, "stability" asks: "If my measurements have a tiny bit of error, how much will my answer about the inside change?"

Good Stability: A tiny error in measurement leads to a tiny error in the answer.
Bad Stability: A tiny error in measurement leads to a huge, wild error in the answer.

The paper shows that for this specific problem, the stability is logarithmic.

The Metaphor: Imagine trying to guess the temperature of a room by looking at a thermometer that is very far away. If you move the thermometer one millimeter, your guess might change by a degree. If you move it two millimeters, your guess might change by ten degrees. The relationship is weak.
The Result: The authors prove that even with this "weak" connection, you can still recover the hidden forces, provided you have some prior knowledge (you know the material near the edge) and you assume the hidden forces aren't infinitely wild (they have a "smoothness" limit).

The Two Different Rules for the Two Forces

The paper finds that the two invisible forces behave differently:

The "Wind" (Vector Field A):
- The authors prove a Log-Type estimate.
- Analogy: If you make a small mistake in your surface measurement, your guess about the "wind" inside gets worse, but only at a rate related to the logarithm of the error. It's bad, but manageable.
The "Pressure" (Function q):
- The authors prove a Log-Log-Type estimate.
- Analogy: This is even trickier. If you make a small mistake in measurement, your guess about the "pressure" gets worse at the rate of the logarithm of the logarithm. This is a very slow, very weak connection. It means the problem is extremely sensitive; tiny measurement errors can lead to massive uncertainty about the pressure inside.

However, there is a twist: If there is no "wind" (A = 0), the problem for the "pressure" (q) gets easier. The stability improves from "Log-Log" to just "Log." It's still unstable, but not as unstable.

How Did They Do It? (The Detective's Toolkit)

To solve this, the authors used a sophisticated mathematical technique involving Complex Geometric Optics (CGO) solutions.

The Metaphor: Imagine shining a very special, invisible laser beam through the block. This beam is designed to oscillate wildly and decay quickly, allowing it to "probe" the deep interior without being blocked by the surface.
By creating two different "beams" (one for the wind, one for the pressure) and seeing how they interfere with each other inside the block, the authors could translate the surface measurements into a map of the interior.
They also used a technique called Carleman Estimates, which acts like a mathematical magnifying glass, proving that if the solution is zero on a small patch, it must be zero everywhere (or very close to it), allowing them to bridge the gap between the tiny measured patch and the rest of the object.

Summary

This paper is a rigorous proof that you can identify hidden forces inside a 3D object by only measuring a tiny part of its surface, even though the math says the answer is very sensitive to measurement errors.

If you know the edge: You can solve the puzzle.
The catch: The solution is "logarithmically unstable," meaning you need extremely precise measurements to get a decent answer.
The difference: Finding the "wind" is slightly easier than finding the "pressure," unless there is no wind at all, in which case finding the pressure becomes slightly easier.

The authors did not claim this can be used in hospitals or for non-destructive testing yet; they simply proved that the mathematical foundation for such a possibility exists, even under the difficult condition of having only partial data.

Technical Summary: Stable Determination of the First Order Perturbation of the Biharmonic Operator from Partial Data

Problem Statement
This paper addresses an inverse boundary value problem for the perturbed biharmonic operator in a bounded domain $\Omega \subseteq \mathbb{R}^n$ ( $n \ge 3$ ) with a smooth boundary $\partial\Omega$ . The operator is defined as $L_{A,q} = \Delta^2 + A(x) \cdot D + q(x)$ , where $D = i^{-1}\nabla$ , $A$ is a vector field in $W^{1,\infty}(\Omega, \mathbb{C}^n)$ , and $q$ is a potential in $L^\infty(\Omega, \mathbb{C})$ . The study focuses on the partial data setting: recovering the coefficients $A$ and $q$ from measurements taken only on arbitrarily small open subsets $\gamma_1, \gamma_2 \subset \partial\Omega$ .

The problem assumes that the coefficients are a priori known in a neighborhood $\omega_0$ of the boundary. The data is provided by a partial Dirichlet-to-Neumann (DtN) map $\Lambda_{\gamma_1, \gamma_2}^{A,q}$ , which maps boundary data $(f, g)$ supported on $\gamma_1$ to the normal derivatives $(\partial_\nu u, \partial_\nu(\Delta u))$ measured on $\gamma_2$ . The primary goal is to establish stability estimates (quantifying how errors in measurements affect the reconstruction of coefficients) for both the first-order perturbation $A$ and the zeroth-order perturbation $q$ .

Methodology
The authors employ the method of Complex Geometric Optics (CGO) solutions, a standard technique in inverse problems for elliptic operators. The proof strategy involves several key technical components:

Construction of CGO Solutions: The authors construct solutions $u_1$ and $u_2$ to the adjoint and forward equations, respectively, of the form $u(x) = e^{ix\cdot\zeta/h}(a(x) + r(x))$ , where $\zeta$ is a complex vector satisfying $\zeta \cdot \zeta = 0$ , $h$ is a semiclassical parameter, and $r$ is a remainder term vanishing as $h \to 0$ .
Integral Identity and Green's Formula: By applying Green's formula to the difference of two operators and utilizing the partial DtN map, the authors derive an integral identity relating the difference of coefficients ( $A_1 - A_2$ and $q_1 - q_2$ ) to the difference of the DtN maps.
Local Quantitative Unique Continuation: A crucial innovation in this work is the derivation of a local quantitative unique continuation estimate for solutions to the nonhomogeneous biharmonic equation with homogeneous boundary conditions. This is achieved using Carleman estimates for the semiclassical biharmonic operator $h^4\Delta^2$ . This step allows the authors to control the solution norms specifically over the measurement subset $\gamma_2$ rather than the entire boundary, which is essential for the partial data setting.
Decomposition of Vector Fields: To recover the vector field $A$ from its exterior derivative $dA$ (which is naturally obtained via the Fourier transform of the integral identity), the authors utilize a Helmholtz-type decomposition $A = A_{sol} + \nabla \phi$ . They estimate the solenoidal part $A_{sol}$ using the bound on $dA$ and the potential part $\nabla \phi$ by solving a boundary value problem for $\phi$ and applying Sobolev embedding and interpolation inequalities.
Optimization of Parameters: The stability estimates are derived by optimizing the semiclassical parameter $h$ and the frequency cutoff $\rho$ in the Fourier domain, balancing error terms arising from the CGO remainders and the exponential weights in the Carleman estimates.

Key Contributions and Results
The paper establishes logarithmic-type and double-logarithmic-type stability estimates, which are characteristic of ill-posed inverse problems.

Stability for the First-Order Perturbation ( $A$ ):
Theorem 1.1 establishes a log-type stability estimate for the vector field $A$ . Specifically, the $L^\infty$ norm of the difference $A_1 - A_2$ is bounded by a sum of a power of the DtN difference and a negative power of the logarithm of the DtN difference:
$\|A_1 - A_2\|_{L^\infty} \lesssim \|\Lambda_1 - \Lambda_2\|^{\mu_1} + |\log \|\Lambda_1 - \Lambda_2\||^{-\mu_2}$
This result complements previous uniqueness results by providing a quantitative stability bound.
Stability for the Zeroth-Order Perturbation ( $q$ ) with First-Order Term:
Theorem 1.2 establishes a log-log-type stability estimate for the potential $q$ when the first-order term $A$ is present:
$\|q_1 - q_2\|_{H^{-1}} \lesssim |\log |\log \|\Lambda_1 - \Lambda_2\|||^{-\mu'}$
Corollary 1.3 extends this to an $L^\infty$ estimate under additional regularity assumptions on $q$ .
Stability for the Zeroth-Order Perturbation ( $q$ ) without First-Order Term:
Theorem 1.4 shows that in the absence of the first-order perturbation ( $A=0$ ), the stability for $q$ improves from log-log-type to log-type:
$\|q_1 - q_2\|_{H^{-1}} \lesssim \|\Lambda_1 - \Lambda_2\|^{2/3} + |\log \|\Lambda_1 - \Lambda_2\||^{-2/(n+2)}$
This highlights the specific difficulty introduced by the first-order term in the biharmonic operator.

Significance and Claims
The authors claim that this work provides the first stability results for the first-order perturbation of the biharmonic operator using partial data. While uniqueness for this problem was previously established (citing [42]), the stability analysis is significantly more challenging.

The paper emphasizes two main technical novelties:

Local Unique Continuation: The derivation of a local quantitative unique continuation estimate for higher-order elliptic operators (specifically the biharmonic operator) using Carleman estimates. The authors note this is the first such result for higher-order operators in this specific partial data context.
Handling Partial Data: The ability to derive stability estimates using measurements on arbitrarily small open subsets of the boundary, rather than requiring data on the entire boundary or specific geometric configurations (like flat inaccessible parts).

The authors position their results as a complement to existing uniqueness theorems and a step forward in the stability analysis of higher-order inverse problems, drawing parallels to methods used for the Schrödinger and magnetic Schrödinger operators while addressing the unique challenges posed by the biharmonic operator's structure and the lack of gauge invariance for its first-order perturbation.