Anderson-Gross-Pitaevskii Equation

• The Anderson-Gross-Pitaevskii equation is a nonlinear Schrödinger model that incorporates confining potentials and spatial white noise to describe disordered Bose-Einstein condensates.
• It employs advanced variational methods, renormalization, and spectral analysis to establish the existence, regularity, and stability of standing-wave solutions in one and two dimensions.
• The model provides insights into the complex interplay between nonlinearity and disorder, guiding research on localization phenomena and stability in stochastic quantum systems.

The Anderson-Gross-Pitaevskii equation (AGP) is a nonlinear Schrödinger equation incorporating both a confining potential and a multiplicative spatial white noise, representing a fundamental model for the study of disordered Bose-Einstein condensates under strong random perturbations. This equation generalizes the classical Gross-Pitaevskii framework, introducing stochasticity that induces rich phenomena such as anomalous localization, random spectral properties, and complex stability regimes. The AGP equation is typically posed in spatial dimensions $d = 1$ or $2$ and supports stationary, time-periodic (standing-wave) solutions whose existence, regularity, decay, and orbital stability are governed by intricate variational and analytic structures (Mackowiak, 28 Dec 2025).

1. Formal Definition and Operator Construction

The AGP equation takes the form

$$i\,\partial_t u + H\,u + \xi\,u + \lambda\,|u|^{2\gamma}u = 0, \qquad u(0,x) = u_0(x),$$

where $H = \Delta - |x|^2$ is the Hermite (confining) operator on $\mathbb{R}^d$, $\xi$ is real-valued spatial white noise, $\lambda\in\mathbb{R}$ is the coupling parameter, and $\gamma > 0$ the nonlinearity exponent. In $d = 1$, the operator $A = H + \xi$ is defined via its quadratic form on the Hermite-Sobolev space $\mathcal{D}(\sqrt{|A|})$.

For $d = 2$, the critical roughness of $\xi \in H^{-1-\epsilon}$ necessitates renormalization via an exponential transform: introducing $Y = (-H)^{-1}\xi$ and $\rho = e^Y$ yields $u = \rho v$ and defines the Wick-renormalized operator

$(H+\xi)u = \rho\bigl[H v + 2\,\nabla Y \cdot \nabla v + x \cdot Y\,x\,v + \wick{|\nabla Y|^2} v\bigr].$

In both cases, $A$ is self-adjoint, lower-semi-bounded, with compact resolvent on $L^2(\mathbb{R}^d)$ (Mackowiak, 28 Dec 2025, Mackowiak, 21 Nov 2025).

2. Standing Waves and the Stationary Elliptic Problem

Standing-wave solutions are sought in the form $u(t,x) = e^{i\omega t} \phi(x)$, yielding the stationary elliptic equation for the profile $\phi$,

$$-A\phi - \lambda |\phi|^{2\gamma} \phi + \omega \phi = 0, \qquad \phi \in \mathcal{D}(\sqrt{|A|}).$$

This equation governs the critical points of the AGP action functional and represents the Euler-Lagrange equation associated with the energy and mass constraints. The variational structure is anchored in the energy space $H^1_A = \mathcal{D}(\sqrt{|A|})$, and standing waves are constructed as minimizers of the constrained energy (Mackowiak, 28 Dec 2025).

3. Existence and Classification of Standing-Wave Solutions

Denote $\mu_0 > 0$ the lowest eigenvalue of $-A$. Existence of positive (nonnegative) standing-wave profiles $\phi \geq 0$ is characterized precisely by the following three regimes:

• $\lambda > 0$ and $\omega > -\mu_0$
• $\lambda = 0$ and $\omega = -\mu_0$
• $\lambda < 0$ and $\omega < -\mu_0$

In all other parameter regimes, no nontrivial nonnegative solution exists in $\mathcal{D}(\sqrt{|A|})$. The proof employs construction of $A$ via quadratic-form techniques (Sturm–Liouville in $d = 1$, exponential transform and Wick renormalization in $d = 2$), variational minimization, and concentration–compactness arguments to secure existence, positivity, and uniqueness (modulo phase) due to the spectral gap of $-A$ (Mackowiak, 28 Dec 2025).

4. Regularity and Spatial Localization

The spatial white noise $\xi$ being distributional restricts the smoothness of solutions. Almost surely, any solution $\phi \in \mathcal{D}(A)$ exhibits

$$\phi \in \bigcap_{q \in [2, \infty)} H^{2-\frac{d}{2}-\epsilon, q}(\mathbb{R}^d), \quad \forall\, \epsilon > 0,$$

and is in $C^{2-d/2-}(\mathbb{R}^d)$ but not better. Maximal regularity is characterized sharply: should $\phi$ possess $H^{2-d/2,q}(U)$ regularity on any open set $U$, it must vanish identically on $U$ (Mackowiak, 28 Dec 2025).

Despite loss of translation invariance by noise, all standing-wave profiles in the focusing regime ($\lambda > 0$) decay exponentially: $$|\phi(x)| + |\nabla\phi(x)| \leq C e^{-c|x|^2}, \qquad x \in \mathbb{R}^d,$$ proven via weighted-multiplier techniques absorbing random potentials as lower-order perturbations (Mackowiak, 28 Dec 2025).

5. Orbital Stability and Dynamical Properties

In the defocusing regime ($\lambda < 0$), the energy-ground-state manifold

$$\left\{e^{i\theta}\phi_m : \theta \in [0,2\pi)\right\} = \left\{u \in H^1_A : \mathcal{M}(u) = m, \mathcal{E}(u) = E(m)\right\},$$

is orbitally stable under the time evolution of the AGP equation. Initial data close (in $H^1_A$) to $\phi_m$ remains close up to phase for all $t$. Stability relies on conservation of mass and energy, coercivity of the second variation of energy at $\phi_m$, and uniqueness/positivity properties of the minimizer (Mackowiak, 28 Dec 2025).

6. Renormalized Functional Spaces and Well-Posedness

In $d = 2$, the operator $A$ is constructed via enhanced noise and gauge transforms. Setting $Y = (-H)^{-1}\xi$, $\rho = e^Y$, one works in the domain $D^{1,2} = \{u = \rho v : v \in H^{1,2}\}$. Wick renormalization and paracontrolled calculus enable analysis despite critical roughness of $\xi$. Sharp Sobolev/Besov space characterizations relate the regularity of $u$ to fractional powers of $A$ (Mackowiak, 21 Nov 2025).

Local well-posedness of the AGP flow is established by contraction in Banach spaces $$X_T = C([0,T];\mathcal{D}^{\alpha,2}) \cap L^p([0,T];L^\infty)$$, provided initial data is taken from suitable weighted Hermite–Sobolev classes $\mathcal{D}^{\alpha,2}$. The solution map is continuous, and maximal solutions either exist globally (for defocusing/subcritical regimes) or blow up in finite time (Mackowiak, 21 Nov 2025).

7. Context: Disordered Lattices, Localization, and Extensions

In lattice models, the Anderson-Gross-Pitaevskii equation generalizes to

$$\left[-J\Delta + \epsilon_n + g |\psi_n|^2\right] \psi_n = \mu \psi_n$$

where $J$ is hopping amplitude and $\epsilon_n$ i.i.d. site potential. Weak interaction ($ga \ll W$) produces Lifshits-localized ground states; strong interaction ($ga \gg W$) yields nearly homogeneous condensates. Linearized excitations (via Bogoliubov-de Gennes analysis) display Anderson localization, with a divergence in localization length for Goldstone modes ($\omega \to 0$) and side-peak singularities at energy $\omega_c \approx 2\sqrt{gaJ}$ due to interaction-induced correlated disorder. This complex interplay is under current analytical and numerical study (Kati et al., 2021).

Summary Table: Key Regimes for Standing-Wave Existence

Coupling $\lambda$	Frequency $\omega$	Existence of Solution
$\lambda > 0$	$\omega > -\mu_0$	Yes
$\lambda = 0$	$\omega = -\mu_0$	Yes
$\lambda < 0$	$\omega < -\mu_0$	Yes
otherwise	otherwise	No

The Anderson-Gross-Pitaevskii equation thus provides a mathematically rigorous and physically relevant framework for understanding ground-state structure, excitation localization, stability, and variational mechanisms in nonlinear quantum systems subject to random environments. These results substantiate the rich spectrum of phenomena emerging from nonlinearity-disorder interplay and guide ongoing studies in stochastic quantum condensates and disordered many-body theories (Mackowiak, 28 Dec 2025, Mackowiak, 21 Nov 2025, Kati et al., 2021).