Dynamics of focusing nonlinear Schrödinger equation with partial harmonic confinement in higher dimensions

1. Problem Statement

The paper investigates the long-time dynamics of the focusing intercritical Nonlinear Schrödinger Equation (NLS) with a partial harmonic confinement in higher dimensions. The equation is given by:
$$i\partial_t u + \Delta_z u - y^2 u = -|u|^\alpha u, \quad z = (x, y) \in \mathbb{R}^d \times \mathbb{R},$$
where $d \ge 3$ is an integer, and the nonlinearity exponent $\alpha$ satisfies the intercritical condition:
$$\frac{4}{d} < \alpha < \frac{4}{d-1} \quad (\text{for } d \ge 2).$$

Key Physical Context:

• Partial Confinement: The potential $y^2$ acts only in the $y$-direction (1D), while the $x$-direction ($\mathbb{R}^d$) remains free. This models Bose-Einstein condensates in anisotropic traps.
• Dispersion vs. Confinement: The $x$-direction provides global dispersion (allowing for scattering), while the $y$-direction is confined (preventing dispersion in that direction).
• The Challenge: Previous work by Ardila and Carles (2021) established a sharp dichotomy between global existence/scattering and blow-up for dimensions $d \le 4$ using the concentration-compactness and rigidity method. However, this method fails for $d \ge 5$ because the nonlinearity $|u|^\alpha$ loses Lipschitz continuity in higher dimensions, breaking the necessary stability estimates.

Objective: To extend the sharp scattering/blow-up dichotomy to all dimensions $d \ge 3$, bypassing the limitations of the concentration-compactness principle.

2. Methodology

The authors introduce a novel strategy that circumvents the concentration-compactness argument, which is typically required for high-dimensional NLS problems. Instead, they rely on interaction Morawetz estimates combined with variational analysis.

A. Variational Characterization and Ground States

• Semivirial Functional: They define a "semivirial" functional $Q(u)$ focusing only on the dispersive $x$-direction:
  $$Q(u) = \|\nabla_x u\|_{L^2}^2 - \frac{\alpha d}{2(\alpha+2)} \|u\|_{L^{\alpha+2}}^{\alpha+2}.$$
  Unlike the standard virial identity, this functional ignores the $y$-derivatives, which is crucial because the $y$-direction does not disperse.
• Minimization Problem: They define a threshold energy level $m_\omega$ via a constrained minimization problem:
  $$m_\omega = \inf \{ S_\omega(\phi) : \phi \in \Sigma \setminus \{0\}, Q(\phi) = 0 \},$$
  where $S_\omega$ is the action functional.
• Key Variational Result: A major technical hurdle is that the constraint $Q(\phi)=0$ is not a natural constraint for the elliptic equation associated with the ground state. The authors prove that the minimizer $\phi_\omega$ of this problem actually satisfies the stationary equation and that $m_\omega$ coincides with the standard ground state energy $m_{GS}$ (the energy of the ground state of the elliptic problem). This is achieved using a truncation semivirial functional and a deformation argument.

B. Interaction Morawetz-Dodson-Murphy (IMDM) Estimates

• Scattering Criterion: Following the roadmap of Dodson and Murphy (2017), the authors establish a scattering criterion based on the smallness of the solution in a specific spacetime norm over a long time interval.
• Anisotropic Multiplier: Standard Morawetz estimates rely on spatial translation invariance. Since the partial harmonic potential breaks this symmetry, the authors construct a multiplier that depends only on the $x$-variable.
• Coercivity Estimate: A critical component is proving that the semivirial functional $Q$ is coercive on the set of solutions with energy below the threshold. They show that if $S_\omega(u) < m_\omega$ and $Q(u) \ge 0$, then $Q$ controls the gradient norm in the $x$-direction. This allows them to close the Morawetz estimate without needing the full concentration-compactness machinery.

3. Key Contributions

1. Removal of Dimensional Restrictions: The paper successfully extends the sharp scattering result to dimensions $d \ge 5$, a regime previously inaccessible due to the lack of Lipschitz continuity in the nonlinearity for the concentration-compactness method.
2. Alternative to Concentration-Compactness: It demonstrates that for NLS with partial harmonic confinement, the interaction Morawetz estimate combined with a variational characterization of the ground state is a more robust tool than the Kenig-Merle concentration-compactness/rigidity argument.
3. Resolution of the Variational Constraint: The authors solve the non-trivial problem of proving that the minimizer of the constrained problem ($Q=0$) is indeed a solution to the elliptic equation, despite the lack of scaling invariance.
4. Handling Anisotropy: The work develops specific tools to handle the anisotropic nature of the problem (different behavior in $x$ vs. $y$), particularly in the construction of the Morawetz multiplier and the proof of coercivity.

4. Main Results

The paper establishes a complete dichotomy classification for the global dynamics of solutions with initial data $u_0$ in the energy space $\Sigma = H^1(\mathbb{R}^{d+1}) \cap L^2(y^2 dz)$.

Let $m_{GS}$ be the ground state energy threshold. The dynamics are determined by the sign of the semivirial functional $Q(u_0)$:

1. Global Existence and Scattering:
   If $S_\omega(u_0) < m_{GS}$ and $Q(u_0) \ge 0$, the solution $u(t)$ exists globally in time and scatters. Specifically, there exist asymptotic states $\phi_\pm \in \Sigma$ such that:
   $$\lim_{t \to \pm \infty} \| u(t) - e^{it(\Delta_z - y^2)} \phi_\pm \|_\Sigma = 0.$$

2. Blow-up:
   If $S_\omega(u_0) < m_{GS}$ and $Q(u_0) < 0$, the solution blows up. This can occur in finite time ($T_{max} < \infty$) or infinite time ($T_{max} = \infty$ but $\|\nabla u(t)\|_{L^2} \to \infty$). If the initial data has finite variance ($|z|u_0 \in L^2$), the blow-up is guaranteed to be in finite time.

Note: The threshold $m_{GS}$ is shown to be equivalent to the variational threshold $m_\omega$ defined by the constrained minimization, ensuring the classification is sharp.

5. Significance

• Theoretical Advancement: This work resolves a significant open problem in the analysis of NLS with partial harmonic oscillators. It proves that the "concentration-compactness" barrier in high dimensions can be overcome for this specific class of equations using Morawetz estimates.
• Methodological Innovation: The paper provides a blueprint for handling NLS equations with anisotropic potentials where standard scaling and translation invariances are broken. The use of the "semivirial" functional to isolate the dispersive directions is a powerful new technique.
• Physical Relevance: The results provide a rigorous mathematical foundation for understanding the stability and long-term behavior of Bose-Einstein condensates in anisotropic traps, specifically clarifying the conditions under which the condensate disperses versus collapses.
• Broader Impact: The techniques developed here (specifically the interaction Morawetz estimates adapted to non-translation-invariant settings) may be applicable to other dispersive equations with partial confinement or on waveguide manifolds.