Can static boundary measurements determine an elastic body’s two moduli?

← All digests

Family 372 · Partial differential equations
Importance 5/5 · Provisional editorial rating
Sources consulted 2026-10-26

1. The question

Suppose an elastic object has unknown stiffness inside. You prescribe a small displacement on its surface and measure the force required to hold that displacement. Repeating this for every possible boundary displacement gives a complete static boundary response.

Can those measurements determine both interior elastic moduli? In three-dimensional isotropic elasticity, one modulus controls shear and another controls compression. Their spatial variation is coupled through a vector-valued equilibrium equation, so recovering them is more complicated than recovering a single conductivity from voltage measurements.

Family 372 claims that the full static boundary map determines both smooth Lamé moduli on every bounded connected smooth domain in $\mathbb R^3$, under positive shear and bulk modulus assumptions. No small variation from constant coefficients, real analyticity or prior agreement near the boundary is required.

This is a uniqueness theorem for exact complete data in a fixed Euclidean domain. It does not supply a practical reconstruction algorithm or a bound for errors caused by measurement noise.

2. Background and essential definitions

Displacement, strain and stress

Let $u:\Omega\to\mathbb R^3$ be the displacement of an elastic body. Its derivative matrix $\nabla u$ contains both deformation and infinitesimal rotation. The symmetric part

$$e(u)=\frac{\nabla u+(\nabla u)^T}{2}$$

is the linearized strain. Taking the symmetric part removes pure infinitesimal rotation, which should not cost elastic energy.

An isotropic material has the same local response in every direction. Its stress tensor is determined by two scalar functions, the Lamé moduli $\lambda(x)$ and $\mu(x)$:

$$\sigma_{\lambda,\mu}(u) =\lambda(\operatorname{div}u)I+2\mu e(u).$$

Here $I$ is the identity matrix and $\operatorname{div}u=\operatorname{tr}e(u)$ measures first-order volume change. The shear modulus is $\mu$; the bulk modulus is $\lambda+2\mu/3$.

Static equilibrium without body forces means

$$\operatorname{div}\sigma_{\lambda,\mu}(u)=0.$$

The divergence acts on the rows of the stress tensor. This is the Lamé system, a coupled elliptic PDE for the three components of $u$.

Why the positivity assumptions have this form

The energy density is

$$\lambda(\operatorname{tr}e)^2+2\mu|e|^2.$$

Decompose strain into its trace-free part $e_0$ and its pure-volume part:

$$e=e_0+\frac{\operatorname{tr}e}{3}I.$$

Then the same density becomes

$$2\mu|e_0|^2+ \left(\lambda+\frac{2\mu}{3}\right)(\operatorname{tr}e)^2.$$

Thus the assumptions

$$\mu>0,\qquad 3\lambda+2\mu>0$$

give positive energy for both shear and compression. The coefficient $\lambda$ itself need not be positive. For smooth coefficients on the compact closure, strict positivity gives uniform positive lower bounds.

The measured operator

Prescribe boundary displacement $f$, and let $u_f$ be the equilibrium solution with that boundary value. The measured traction is the force per unit area,

$$\Lambda_{\lambda,\mu}f =\sigma_{\lambda,\mu}(u_f)n,$$

where $n$ is the outward normal. The displacement-to-traction map $\Lambda$ records this response for every $f$.

Its weak form is

$$\langle\Lambda f,g\rangle =\int_\Omega \bigl[\lambda(\operatorname{div}u_f)(\operatorname{div}v_g) +2\mu e(u_f):e(v_g)\bigr],dx,$$

where $v_g$ is any Sobolev extension of the boundary displacement $g$, and $A:B=\sum_{i,j}A_{ij}B_{ij}$. Equilibrium makes the value independent of the chosen extension.

The natural trace spaces are $H^{1/2}$ for displacement and $H^{-1/2}$ for traction. They allow the boundary work pairing to remain meaningful when the data are less regular than classical smooth functions.

3. Development history and previous partial results

Calderón’s conductivity problem and the Sylvester–Uhlmann complex geometric optics method established the broader inverse-boundary framework. Elasticity adds two coefficient fields and vector coupling; scalar uniqueness cannot simply be applied component by component.

Nakamura and Uhlmann recovered the boundary Taylor series of the elastic coefficients. Their 1994 global uniqueness claim was subsequently corrected by a 2003 erratum. The corrected theorem assumes sufficiently small derivatives of the shear moduli in a suitable smooth norm.

The erratum identifies difficulties in a prescribed planar matrix transport problem and in the type of equation obtained from corrected complex geometric optics solutions. This is part of the mathematical development history: the unrestricted smooth three-dimensional theorem should not be attributed to the original claim without its correction.

Eskin and Ralston developed matrix complex geometric optics expansions and restricted elasticity uniqueness results. Their auxiliary vector–scalar reductions are useful, but an auxiliary field can contain information not determined by the corresponding physical displacement. Equality of physical boundary measurements must therefore be transferred with care.

In two dimensions, Imanuvilov and Yamamoto proved global uniqueness for smooth Lamé coefficients under the appropriate positivity assumptions, without a smallness condition. The different dimension changes the available analytic methods.

For three dimensions, Lin and Nakamura developed boundary reconstruction at finite regularity. Tan and Liu recovered boundary jets on smooth Riemannian manifolds and obtained global recovery under analytic hypotheses. Recent work by Chen, Jiang, Liu and Tao treats structured shear moduli under specified symmetry, separation or quasianalyticity assumptions.

The new manuscript claims global uniqueness for the unrestricted smooth isotropic class under positive elastic energy. Its method combines physical displacement transfer, compatible complex geometric optics solutions and holomorphic dependence on complex directions.

4. What the new manuscript claims

Let $\Omega\subset\mathbb R^3$ be any bounded connected domain with $C^\infty$ boundary. For $j=1,2$, let $\lambda_j,\mu_j$ be real functions smooth on the closure and satisfying

$$\mu_j>0,\qquad3\lambda_j+2\mu_j>0 \quad\text{on }\overline\Omega.$$

The theorem states

$$\Lambda_{\lambda_1,\mu_1}=\Lambda_{\lambda_2,\mu_2} \quad\Longrightarrow\quad \lambda_1=\lambda_2,\qquad\mu_1=\mu_2 \quad\text{throughout }\Omega.$$

The measured map is the full zero-frequency displacement-to-traction operator. Both inputs and observations range over the whole boundary. Unlike the local conductivity claims discussed in family 365, this theorem is not stated for one arbitrary boundary patch.

The domain and Euclidean coordinates are fixed. The conclusion is exact equality of the two scalar coefficient fields, not equality modulo an arbitrary interior coordinate transformation.

No simple-connectivity assumption is imposed. The boundary may have several components. The moduli need not be close to constants, and agreement in a boundary neighborhood is not an input hypothesis: the source derives the required matching boundary information from the measurements.

5. Why the advance matters

Importance: 5/5. If correct, the manuscript resolves a central smooth inverse problem for three-dimensional isotropic elasticity. It establishes that the full static response contains enough information to separate and recover both shear and compressional behavior.

The result concerns identifiability before computation. If two different admissible interiors could produce identical complete data, no algorithm could distinguish them from those data alone. The claimed theorem excludes that ambiguity in the stated class.

Its proposed method also addresses a recurring issue in inverse systems: an analytically convenient augmented equation may have more solutions than the physical model. Recovery must use the boundary information actually available for physical displacements, not stronger measurements of auxiliary variables that were never prescribed.

6. The main idea

Transfer genuine displacements between the two media

Equality of boundary maps first gives equality of all boundary derivatives of the moduli. The manuscript extends both coefficient pairs smoothly outside $\Omega$, making them identical there and constant sufficiently far away.

Take an exact equilibrium displacement in the first extended medium. In the second medium, solve the interior problem with the same boundary displacement and glue it to the first solution outside. Equality of traction cancels the interface term in the weak equation, so the glued field is an exact physical solution of the second system.

The two displacements and their actual divergences agree outside the object. This exterior agreement is the starting information for the analytic comparison; it is obtained from the physical map itself.

Add divergence without forgetting its meaning

Adjoining a scalar variable $d$, intended to equal $\operatorname{div}u$, turns the system into an augmented vector–scalar equation whose highest-order part is the ordinary Laplacian on each component.

That simplification is useful for estimates, but arbitrary augmented solutions need not satisfy $d=\operatorname{div}u$. The manuscript retains a compatibility identity for this difference and constructs exact physical displacements after correcting approximate solutions. Thus the scalar component remains the true divergence in the comparisons that use measured data.

The normalized augmented operators for two coefficient pairs have the same highest-order part. Their difference contains only first derivatives and zeroth-order terms, which is crucial for controlling transferred solutions.

Probe with complex exponential solutions

Choose a nonzero complex vector $\theta$ with

$$\theta\cdot\theta=0.$$

The dot product here is complex bilinear, without complex conjugation. Such vectors exist: $(1,i,0)$ is an example. The null condition makes the largest Laplacian term vanish when testing with an exponential $e^{\tau\theta\cdot x}$, leaving transport equations for its amplitude.

The real and imaginary parts of $\theta$ span a plane. On that plane, $D_\theta=\theta\cdot\nabla$ acts like a complex Cauchy–Riemann derivative. Leading displacement amplitudes $a$ satisfy $\theta\cdot a=0$; together with a scalar divergence amplitude they form a three-dimensional complex space.

The manuscript constructs a full local frame in this constrained space using exact physical solutions. High-order transport expansions and weighted estimates control both the leading amplitudes and their first derivatives. This is stronger than constructing formal auxiliary solutions with no physical compatibility.

Compare transport frames as direction varies

After an invertible normalization, each medium’s leading transport has a universal upper equation of the form

$$D_\theta P=\theta s,$$

where $P\in\theta^\perp$ and $s$ is scalar. The coefficient dependence enters the remaining equation.

Transfer a first-medium frame through the physical solution construction. Limits of the transferred amplitudes produce a comparison matrix $G$, equal to the identity outside a fixed ball. Only the original frame must be invertible; the argument does not initially assume that the transferred columns form a frame.

Supported transport uniqueness joins the local comparisons and makes them vary holomorphically with the complex direction. Directions are considered up to nonzero complex multiples; the null directions form a compact complex curve, the projective null conic.

The spaces $\theta^\perp$ vary with direction and form a vector bundle: a smoothly and holomorphically organized family of vector spaces. The manuscript uses this bundle’s restricted holomorphic sections, together with the universal upper transport equation, to force $G$ to be the identity. A rank comparison on the two-dimensional space $\theta^\perp$ removes the remaining off-diagonal freedom.

Recover the moduli from identical transports

Equality of transport coefficients first determines the ratio

$$\frac{\lambda+\mu}{\lambda+2\mu}.$$

The remaining difference is expressed through

$$\eta=\log\mu_1-\log\mu_2.$$

The manuscript derives a system of the form

$$\nabla^2\eta=B(x)\nabla\eta+q(x)I,$$

where $B(x)$ is a known smooth linear map from vectors to matrices and $q$ is an auxiliary scalar. Differentiating and commuting mixed derivatives closes a homogeneous first-order system for $(\nabla\eta,q)$.

These quantities vanish in the common exterior. Along any path the system becomes a linear ordinary differential equation with zero initial values, so uniqueness propagates their vanishing throughout the connected region. The exterior normalization fixes the remaining constant in $\eta$, giving $\mu_1=\mu_2$. The already determined ratio then gives $\lambda_1=\lambda_2$.

7. Scope, limitations, and open questions

The theorem concerns smooth real isotropic coefficients in three dimensions with positive shear and bulk moduli. Rough, complex, anisotropic or degenerating coefficients are outside its statement. It also concerns static linearized elasticity, not nonlinear material response or an unknown mass density in a time-dependent problem.

Full boundary data mean an entire operator, not finitely many loading experiments. The result does not quantify reconstruction stability, give a noise tolerance or specify a numerical recovery method.

Partial-boundary recovery, lower regularity and quantitative estimates remain separate questions. The connection to family 365 is methodological, through complex transport and inverse-boundary geometry; its different operator and measurement assumptions should be retained.

8. Sources and further reading

The Nakamura–Uhlmann erratum is essential to the historical account. The Imanuvilov–Yamamoto, Lin–Nakamura and Tan–Liu papers linked above describe the two-dimensional and boundary-determination precedents. Eskin–Ralston supply important matrix complex geometric optics methods.

For optional background, Lamé parameters and linear elasticity explain the material interpretation. The manuscript below is the source of the unrestricted smooth three-dimensional uniqueness claim.

Source manuscripts

• M1. OpenAI, Global Uniqueness for the Smooth Isotropic Elasticity Inverse Problem (September 24, 2026).

Sources consulted on 26 October 2026 at repository revision 1581d4c. Download the manuscript citations.