Geometric condition for the observability of electromagnetic Schrödinger operators on $\mathbb{T}^2$

Technical Summary: Geometric Condition for the Observability of Electromagnetic Schrödinger Operators on $T^2$

Problem Statement
The paper investigates the observability of the Schrödinger equation on the two-dimensional torus $T^2 = \mathbb{R}^2/(2\pi\mathbb{Z})^2$ in the presence of an electromagnetic field. The system is governed by the operator $H_{A,V} = (\frac{1}{i}\nabla - A(z))^2 + V(z)$, where $A$ is a smooth magnetic potential and $V$ is an electric potential. The central question is whether the $L^2$-norm of the initial state $u_0$ can be estimated by the $L^2$-norm of the solution restricted to a non-empty open observation set $\omega \subset T^2$ over a time interval $(0, T)$.

While it is well-established that for the pure electric case ($A \equiv 0$), observability holds for any non-empty open set $\omega$ (regardless of whether it satisfies the Geometric Control Condition, GCC), the authors demonstrate that the presence of a non-constant magnetic potential introduces a new obstruction. Specifically, if $\omega$ does not satisfy the classical GCC (i.e., there exist closed geodesics avoiding $\omega$), the magnetic field can prevent observability unless specific geometric conditions are met regarding the critical points of the magnetic potential.

Methodology
The authors employ a semiclassical analysis approach, reducing the observability problem to the study of high-frequency solutions (semiclassical limit $h \to 0$). The proof strategy involves several key technical steps:

1. Semiclassical Reduction: The problem is reduced to proving a sharp semiclassical observability estimate for solutions concentrated on the energy shell $|\zeta|=1$. This involves analyzing semiclassical defect measures (Wigner measures) associated with sequences of solutions.
2. Microlocal Analysis and Propagation: The support of the semiclassical measure is shown to be invariant under the geodesic flow. The authors first eliminate irrational directions (where trajectories are dense) and reduce the problem to a finite number of periodic directions.
3. Normal Form Reduction: For a specific periodic direction $\zeta_0 = (1,0)$, the authors perform a two-step normal form reduction (conjugation by unitary operators) to simplify the magnetic Schrödinger operator.
   • First Averaging: A gauge transformation removes the $x$-dependence of the first component of the averaged magnetic potential, $A_1$.
   • Second Averaging: A second conjugation removes the $x$-dependence of the second component, effectively averaging the magnetic potential along the trapped direction.
4. Frequency Localization: The solution is decomposed into transversal high-frequency and low-frequency parts relative to the direction of the geodesic.
   • High-Frequency Part: Analyzed using commutator methods and energy estimates, showing that these components are observable due to their rapid oscillation.
   • Low-Frequency Part: Reduced to a one-dimensional model equation involving an anharmonic oscillator (in the non-degenerate case) or a harmonic oscillator. The analysis relies on the concentration of solutions near the critical points of the averaged magnetic potential.
5. Optimality Construction: To prove the necessity of the condition, the authors construct quasimodes (approximate solutions) concentrated near a non-degenerate critical point of the magnetic potential that lies outside the observation set, demonstrating that observability fails in this scenario.

Key Contributions and Results

• Magnetic Geometric Control Condition (MGCC): The authors introduce a new geometric condition, the MGCC, which is both sufficient and almost necessary for observability.
  • Definition: For every periodic direction $\vec{\gamma}$, the projection of the observation set $\omega$ onto the direction orthogonal to $\vec{\gamma}$ must contain all critical points of the averaged magnetic potential $A_\gamma = \langle A \rangle_{\vec{\gamma}} \cdot \vec{\gamma}$.
  • Equivalence: This is equivalent to requiring that the projection contains all zeros of the averaged magnetic field $\langle B \rangle_\gamma$.
• Main Theorem (Theorem 1.4): If the MGCC is satisfied, then the electromagnetic Schrödinger equation is observable from $\omega$ for any time $T > 0$. This result holds for smooth potentials $A$ and $V$.
• Resolvent Estimates (Theorem 1.6): The paper establishes a resolvent estimate for the stationary operator $H_{A,V}$ under the MGCC, which implies observability even for time sets of positive Lebesgue measure (via recent results by Burq and Zhu).
• Optimality (Theorem 1.8): The MGCC is shown to be sharp. If there exists a periodic direction $\vec{\gamma}$ such that the averaged potential $A_\gamma$ has a non-degenerate critical point outside the projection of $\omega$, then observability fails for any time $T$.
• Control and Stability: As direct consequences of the observability estimates, the authors deduce:
  • Exact Controllability: The controlled Schrödinger equation is exactly controllable in any time $T > 0$ if MGCC holds.
  • Exponential Stability: The damped Schrödinger equation exhibits exponential decay if the damping region satisfies MGCC.

Significance and Claims
The paper claims to resolve the open problem of observability for the electromagnetic Schrödinger equation on $T^2$ when the observation set does not satisfy the classical Geometric Control Condition. The authors emphasize that the magnetic potential fundamentally alters the observability landscape compared to the pure electric case.

• Distinction from Pure Electric Case: Unlike the electric case where any open set suffices, the magnetic case requires the observation set to "capture" the critical points of the magnetic potential along periodic trajectories.
• Geometric Interpretation: The condition is intrinsic to the magnetic field, linking the geometry of the observation set to the topology of the magnetic field's critical points.
• Relation to Quantum Unique Ergodicity (QUE): The authors note a connection to recent work on Quantum Unique Ergodicity for magnetic Laplacians (Morin and Rivière). While the conditions differ (QUE requires non-vanishing averaged fields, which is impossible on $T^2$ due to flux constraints), the techniques and the resulting resolvent estimates are related. The authors suggest their results could likely extend to general magnetic Laplacians on line bundles.

The paper does not propose experimental applications or future directions beyond the mathematical framework established, maintaining a focus on the rigorous derivation of the geometric condition and the associated spectral and control-theoretic estimates.