On the Well-posedness of Magnetic Schrödinger Equations with Unbounded Potentials

1. Problem Statement

The paper addresses the global well-posedness of the Cauchy problem for the magnetic Schrödinger equation with unbounded electromagnetic potentials on $\mathbb{R}^d$ ($d \geq 2$). The equation is given by:
$$i\partial_t u(t, y) = (H_A + V(t))u(t, y), \quad u(0) = u_0$$
where:

• $H_A = \frac{1}{2}\sum_{j=1}^d (-i\partial_j - A_j(y))^2$ is the magnetic Laplacian with a vector potential $A$.
• $V(t, y)$ is a time-dependent electric scalar potential.
• Unboundedness: The potentials are allowed to grow at infinity: $A$ has at most linear growth, and $V$ has at most quadratic growth.
• Regularity: The magnetic field $B = dA$ is smooth with specific decay conditions on its derivatives, while $V$ is smooth with uniform bounds on derivatives of order $\geq 2$.

The Challenge:
Previous results for such unbounded potentials were largely restricted to $L^2$-based Sobolev spaces (e.g., Yajima [18]). Standard $L^p$ spaces ($p \neq 2$) are unsuitable because the free Schrödinger group $e^{-it\Delta}$ is unbounded on $L^p$. While modulation spaces $M^{p,q}$ have successfully replaced $L^p$ for non-magnetic equations, extending these results to the magnetic case with unbounded potentials requires a new framework that respects the gauge invariance and the specific phase-space structure induced by the magnetic field.

2. Methodology

The authors employ a phase-space approach combining magnetic pseudo-differential calculus with magnetic modulation spaces.

A. Magnetic Pseudo-Differential Operators (ΨDOs)

Instead of treating the magnetic potential as a perturbation, the authors utilize the magnetic Weyl quantization developed by Iftimie, Măntoiu, and Purice.

• The operator is defined via a symbol $h(y, \eta)$ and a magnetic phase factor $e^{-i\phi_A(y', y)}$:
  $$\text{Op}_A(h)u(y) = (2\pi)^{-d} \int_{\mathbb{R}^{2d}} e^{i(y-y')\cdot\eta} e^{-i\phi_A(y',y)} h\left(\frac{y+y'}{2}, \eta\right) u(y') dy' d\eta$$
• The phase $\phi_A$ is constructed via the transversal gauge to ensure gauge covariance. This formulation allows the magnetic field $B$ to be the primary object, making the theory intrinsically dependent on $B$ rather than the specific choice of potential $A$.

B. Magnetic Modulation Spaces ($M^p_A$)

The authors define a new function space adapted to the magnetic field:

• Magnetic Wavepacket Transform ($T^A_\lambda$): A generalization of the standard wavepacket transform using magnetic wavepackets $g^A_{x,\xi}$ which include the phase factor $e^{i\phi_A(y,x)}$.
  $$T^A_\lambda u(x, \xi) = \int_{\mathbb{R}^d} e^{-i\xi\cdot(y-x)} e^{-i\phi_A(y,x)} g(\lambda(y-x)) u(y) dy$$
• Space Definition: The magnetic modulation space $M^p_A(\mathbb{R}^d)$ consists of distributions $u$ such that $T^A_\lambda u \in L^p(\mathbb{R}^{2d})$.
• Key Property: These spaces are gauge-independent (isometrically isomorphic for different potentials generating the same $B$) and generalize standard modulation spaces ($M^p_A \cong M^p$ when $A \equiv 0$).

C. Phase-Space Approximation Strategy

The core proof strategy, inspired by Tataru [16] and Yajima [18], involves approximating the quantum evolution by a classical magnetic Hamiltonian flow:

1. Conjugation: The operator $\text{Op}_A(h)$ is conjugated by the wavepacket transform $T^A_\lambda$ to act on phase-space functions.
2. Approximation: The action of $\text{Op}_A(h)$ on a wavepacket is approximated by a magnetic Hamiltonian vector field $\tilde{H}_B$ derived from the symbol $h$ and the magnetic field $B$.
   $$\text{Op}_A(h) g^A_{x,\xi} \approx (i\tilde{H}_B + m_A(h)) g^A_{x,\xi}$$
3. Flow Construction: A parametrix (approximate solution) is constructed using the magnetic Hamiltonian flow $\chi_B(h)$, which solves the classical equations of motion:
   $$\dot{x}_t = \partial_\xi h, \quad \dot{\xi}_t = -\partial_x h + B(x_t) \cdot \partial_\xi h$$
4. Error Control: The difference between the exact quantum evolution and the classical flow approximation is treated as a remainder operator $K(t,s)$. The authors prove that this remainder is a contraction in the appropriate modulation space norms using Volterra equation techniques and specific decay estimates of the integral kernels.

3. Key Contributions

1. Introduction of Magnetic Modulation Spaces ($M^p_A$):
   The paper rigorously defines and analyzes $M^p_A(\mathbb{R}^d)$ and their weighted versions $M^{m,p}_A(\mathbb{R}^d)$. It establishes that these spaces are the natural domain for magnetic Schrödinger operators with unbounded potentials, extending the success of modulation spaces from the non-magnetic case.

2. Gauge Invariance:
   A significant theoretical contribution is the proof that $M^p_A$ depends intrinsically on the magnetic field $B$ rather than the vector potential $A$. This resolves the gauge-dependence issue often found in magnetic problems.

3. Global Well-Posedness for Unbounded Potentials:
   The authors prove global existence and uniqueness of strong solutions for initial data in $M^{2,p}_A(\mathbb{R}^d)$ (which includes $L^2$ and weighted Sobolev spaces) under the assumption of sublinear magnetic potentials and subquadratic electric potentials.

4. Extension to Time-Dependent Fields:
   The methodology is extended to handle time-dependent magnetic fields $B(t)$, proving well-posedness for the Cauchy problem where the magnetic field evolves in time, provided it satisfies specific smoothness and decay conditions.

4. Main Results

Theorem 1.5 (Main Theorem):
Let $T > 0$ and $1 \leq p \leq \infty$. For a magnetic field $B$ satisfying decay conditions and a symbol $h$ (representing the Hamiltonian) satisfying second-order growth conditions:

• The Cauchy problem $i\partial_t u = \text{Op}_A(h)u$ with $u(0) \in M^{2,p}_A(\mathbb{R}^d)$ has a unique strong solution $u \in W^{1,1}([0, T], M^p_A(\mathbb{R}^d))$.
• The solution satisfies the bound:
  $$\|u(t)\|_{M^{2,p}_A} \leq C_{T,h,B} \|u_0\|_{M^{2,p}_A}$$
• This result holds for $p=2$ (recovering $L^2$ results) and $p \neq 2$ (new results).

Theorem 1.7 (Time-Dependent Case):
The well-posedness result extends to the case where the magnetic field $B(t)$ and vector potential $A(t)$ are time-dependent, provided $B(t)$ and $\dot{B}(t)$ satisfy uniform decay and boundedness conditions.

Corollary 1.6:
The result extends to the inhomogeneous equation with a source term $f \in L^1([0, T], M^{s,p}_A)$.

5. Significance

• Bridging the Gap: This work bridges the gap between the well-understood theory of Schrödinger equations in $L^2$ and the more complex behavior in $L^p$ ($p \neq 2$) for magnetic systems. It demonstrates that modulation spaces are the correct setting for analyzing dispersive equations with unbounded potentials.
• Handling Unboundedness: By treating the magnetic potential as part of the principal operator (via magnetic Weyl calculus) rather than a perturbation, the authors overcome the limitations of standard perturbation theory, which fails for unbounded potentials.
• Physical Relevance: The results apply to physical models involving strong, non-uniform magnetic fields (e.g., in plasma physics or condensed matter) where potentials grow at infinity, providing a rigorous mathematical foundation for the stability of quantum states in such environments.
• Methodological Innovation: The combination of magnetic wavepacket transforms with classical Hamiltonian flow approximations offers a robust template for analyzing other quantum systems with geometric or magnetic constraints.

In summary, Frey and Weng successfully establish that magnetic Schrödinger equations with unbounded potentials are globally well-posed in magnetic modulation spaces, utilizing a sophisticated phase-space analysis that respects the underlying gauge geometry of the magnetic field.