Magnetic Dirichlet-to-Neumann Map

• Magnetic Dirichlet-to-Neumann map is a boundary operator defined via modified normal derivatives that encapsulate both magnetic and electric potential effects in PDE analysis.
• It functions as a pseudodifferential operator whose spectral properties yield key geometric invariants and stability estimates vital for inverse problem applications.
• Advanced techniques, including complex geometric optics and Carleman estimates, enable precise reconstruction of interior magnetic and electric potentials from boundary measurements.

The magnetic Dirichlet-to-Neumann (DtN) map is a central boundary operator in the analysis of partial differential equations (PDEs) with magnetic (and often electric) potentials, notably in the context of the magnetic Schrödinger operator. It encodes how boundary input (Dirichlet data) is transported to boundary output (Neumann-type data) under the influence of a magnetic vector potential, reflecting both geometric and topological aspects of the domain, as well as the underlying physical fields. The study of this map spans linear and nonlinear PDEs, both static and dynamical, and extends across Riemannian, complex, and Lorentzian geometric settings.

1. Definition and Classes of Magnetic Dirichlet-to-Neumann Maps

The magnetic DtN map arises from boundary value problems for elliptic (and, in time-dependent settings, hyperbolic) operators modified by a magnetic potential. Let $(M,g)$ be a smooth Riemannian manifold with boundary $\partial M$, and $A$ a (real or complex-valued) smooth 1-form representing the magnetic potential. The prototypical operator is the magnetic Schrödinger operator: $L_{g,A,q}u = (d + iA)^*(d + iA)u + q(x)u$ where $q$ is an electric potential and the adjoint is taken with respect to $g$. For Dirichlet data $f \in H^{1/2}(\partial M)$, the boundary problem

$$L_{g,A,q}\,u = 0 \;\text{in } M, \qquad u|_{\partial M} = f$$

has, under suitable invertibility assumptions, a unique solution $u$. The magnetic normal derivative is defined as

$\partial_\nu^A u := \partial_\nu u + i\,A(\nu)\,u$

with $\nu$ the outward unit normal. The magnetic DtN map is then

$\Lambda_{A,q}(f) = \partial_\nu^A u|_{\partial M}$

yielding a bounded operator $H^{1/2}(\partial M) \to H^{-1/2}(\partial M)$ in the general elliptic case (Liu et al., 2021, Cekić et al., 2024).

In dynamical (parabolic or hyperbolic) settings, or on Lorentzian manifolds, similar constructions hold, with the map defined on traces of solutions to the respective operator equations (Bellassoued, 2015, Stefanov et al., 2016). For nonlinear magnetic Schrödinger equations, where $A$ and $q$ depend also on $u$, the map is nonlinear in general and its definition adapts to the solution structure of the nonlinear problem (Krupchyk et al., 2021, Lai et al., 2020).

For Beltrami fields (eigenfields of the curl operator), a closely related normal-to-tangential map is defined, reflecting the geometric structure of the solution space rather than the potential (Enciso et al., 2024).

2. Analytic and Spectral Properties

The magnetic DtN map is a pseudodifferential operator (ΨDO) of order $1$ (or order $0$ for the Beltrami case) on the boundary. For smooth data, it is self-adjoint and elliptic, and its spectrum—the magnetic Steklov eigenvalues—is discrete and of finite multiplicity (Cekić et al., 2024, Liu et al., 2021). In the presence of magnetic fluxes or for domains with nontrivial topology, the eigenvalue asymptotics and the spectral behavior are sensitive to holonomies of $A$, local fluxes, boundary geometry, and, in the nonlinear case, to the higher-order Taylor coefficients of $A$ and $q$ in $u$ (Cekić et al., 2024, Krupchyk et al., 2021, Bernard et al., 18 Mar 2025).

Explicit diagonalizations exist in symmetric domains (e.g., the disk), where Fourier mode decompositions yield analytically tractable formulas for eigenvalues in terms of special functions (e.g., confluent hypergeometric functions for exterior disk problems (Bernard et al., 18 Mar 2025)). The map's structure is further encapsulated in heat trace expansions, whose coefficients provide spectral invariants encoding boundary geometry, curvature, and the physical fields (Liu et al., 2021).

On Lorentzian manifolds, the map is a Fourier integral operator (FIO) whose canonical relation is dictated by the light-ray (lens) geometry; its symbol structure reveals light ray transforms of the magnetic and electric potentials (Stefanov et al., 2016, Yi et al., 19 May 2025).

3. Inverse Problems, Uniqueness, and Stability Results

A central topic is whether, and to what extent, the magnetic DtN map determines the interior magnetic and electric potentials. In linear settings, only the gauge-invariant content (the magnetic field $dA$ and the electric potential $q$) can be recovered, with $A$ determined only up to an exact form vanishing on the boundary (Bellassoued, 2015, Bellassoued et al., 2016). For example, on simple manifolds and under full or partial boundary data, $dA$ and $q$ are uniquely determined by $\Lambda_{A,q}$, with explicit quantitative Hölder stability estimates (Bellassoued, 2015, Bellassoued et al., 2016, Mejri, 2016, Chung, 2011).

For nonlinear magnetic Schrödinger operators—both in Euclidean domains and on complex Kähler manifolds—the nonlinear DtN map determines both the full (nonlinear) magnetic potential $A(x,u)$ and the electric potential $q(x,u)$ uniquely, with no gauge ambiguity, provided appropriate boundary conditions and holomorphicity assumptions are satisfied. Boundary measurements, including partial data, suffice for this determination (Krupchyk et al., 2021, Lai et al., 2020).

In time-harmonic Maxwell problems, the electromagnetic DtN map is analytic with respect to frequency and encodes material properties and passive system constraints, with Herglotz (Nevanlinna) properties governing spectral representations and positivity (Cassier et al., 2015).

In the context of inverse spectral problems, the Steklov spectrum of the magnetic DtN map can, in favorable situations, uniquely determine boundary invariants such as the number and lengths of boundary components, parallel transport (holonomy), and the magnetic flux along boundaries. However, intricate non-uniqueness phenomena arise due to covering system artifacts in the spectrum, and distinct boundary decompositions may be Steklov-isospectral (Cekić et al., 2024).

On Lorentzian manifolds, knowledge of the magnetic DtN map (even on disjoint source and observation sets) can recover conformal structures, tangential components of $A$, lens relations, and jets of the boundary metric, up to gauge (Stefanov et al., 2016, Yi et al., 19 May 2025).

4. Methods of Analysis: Linearization and Complex Geometric Optics

Integral identities obtained via higher-order expansions (linearizations) of the DtN map with respect to boundary data are essential for uniqueness and reconstruction results, particularly in nonlinear settings (Krupchyk et al., 2021, Lai et al., 2020). In these schemes, the $m$-th order linearization yields integral relations involving Taylor coefficients of the potentials and products of harmonic (or more generally, complex geometric optics) (CGO) solutions. Density results, Runge approximation, and boundary determination lemmas are then employed to conclude uniqueness.

Construction of CGO solutions is fundamental in both linear and nonlinear settings, on Riemannian, complex, or Lorentzian manifolds. These solutions, adapted to the geometry (e.g., holomorphic and anti-holomorphic harmonics on Kähler manifolds), allow one to probe the operator's response to highly oscillatory input and extract integral transforms of the potentials (Krupchyk et al., 2021, Bellassoued, 2015, Cekić et al., 2024, Stefanov et al., 2016). In time-dependent and waveguide problems, geometric optics solutions in the time-frequency domain afford similar access to the coefficients inside the domain (Mejri, 2016, Bellassoued et al., 2016).

In situations involving only partial data, advanced Carleman estimates and microlocal analysis provide control over the boundary propagation of information (Chung, 2011, Stefanov et al., 2016). On Lorentzian spacetime domains, the microlocal structure of the DtN map, as a FIO, encodes the lens relation and integral transforms along light rays (Stefanov et al., 2016, Yi et al., 19 May 2025).

5. Spectral and Geometric Invariants

The spectrum of the magnetic DtN map (magnetic Steklov eigenvalues) encodes a wealth of geometric and physical information. Asymptotic expansions deliver explicit invariants:

• The leading term determines total boundary length or volume.
• Subleading and higher-order terms involve mean curvature, magnetic flux, boundary parallel transport (holonomy), and integrals over curvature and electromagnetic fields (Cekić et al., 2024, Liu et al., 2021).
• In the exterior disk, explicit analytic expressions are available for all spectral branches, capturing flux dependence, diamagnetic monotonicity, and the behavior under strong and weak field asymptotics (Bernard et al., 18 Mar 2025).

Notably, in surfaces with multiple boundary components, the magnetic DtN spectrum may fail to distinguish certain boundary configurations—demonstrating non-uniqueness even to infinite spectral order in the presence of nontrivial magnetic holonomy (Cekić et al., 2024).

For Beltrami fields, the symbol of the normal-to-tangential boundary map determines the full jet of the metric at the boundary, providing a spectrally encoded description of the boundary geometry (Enciso et al., 2024).

6. Applications and Physical Significance

The magnetic DtN map is pivotal in boundary control, nondestructive testing, quantum waveguide analysis, and electromagnetic inverse problems. It is central to the Calderón problem for magnetic Schrödinger and Maxwell operators, identification of material parameters in composites, and the study of quantum systems with nontrivial topology (e.g., Aharonov–Bohm effects). The analytic and spectral structure of the map supports high-precision inverse reconstructions, stability quantifications, and the interpretation of physical measurements in terms of underlying geometry and fields.

Diamagnetic monotonicity and the absence of paramagnetism in the spectral flow of the map as a function of magnetic field strength have concrete implications for transport and transmission properties (Bernard et al., 18 Mar 2025). The entire operator-theoretic framework extends to nonlinear, dynamic, and manifold-based settings, highlighting the universality of the magnetic DtN map as a probe of combined geometry and field content.

7. Challenges, Limitations, and Open Problems

Open challenges include the extension of stability and uniqueness results to lower regularity settings, partial data configurations for nonlinear or anisotropic media, and the classification of spectral non-uniqueness phenomena arising from covering system artifacts or nontrivial holonomy (Cekić et al., 2024, Lai et al., 2020, Mejri, 2016). The classification of all possible groups of isospectral boundary decompositions remains a purely arithmetic and combinatorial frontier.

For time-dependent and Lorentzian manifolds, the fine structure of lens relations, gliding rays, and the subtleties of microlocal gauge fixing lead to complex interplay between analytic, geometric, and physical invariants (Stefanov et al., 2016, Yi et al., 19 May 2025). In the context of Beltrami fields, the lack of Green's functions requires substitutes with controlled analytic properties (Enciso et al., 2024). The full exploitation of higher spectral invariants and the development of constructive reconstruction procedures, especially for nonlinear problems and vector-bundle connections, remain active and challenging areas of contemporary research.