Poisson/Monge-Ampère Coupling Overview

• Poisson/Monge-Ampère coupling is an analytical framework that unites the linear Poisson equation with the nonlinear Monge-Ampère equation to address complex PDE challenges.
• The method employs iterative schemes like the L-scheme, where the Monge-Ampère residual is linearized via Poisson subproblems, ensuring robust convergence even for degenerate or non-convex initial conditions.
• Applications extend to optimal transport, cosmological modeling, and fluid mechanics, demonstrating efficient error control, mesh-independence, and practical numerical realizations.

Poisson/Monge-Ampère coupling refers to analytic and computational frameworks in which the classical linear Poisson equation and the fully nonlinear Monge-Ampère equation are studied together, either through iterative schemes, physical modeling, or geometric analysis. This coupling arises naturally in diverse contexts including numerical solution of PDEs, optimal transport, cosmology, and fluid mechanics, encapsulating linear and nonlinear elliptic phenomena within a unified or iterative pipeline.

1. Formulations of the Poisson and Monge-Ampère Equations

The Poisson and Monge-Ampère equations are prototypical elliptic PDEs, distinguished by linearity and the order of nonlinearity:

• Poisson equation:

$\nabla^2\phi(x) = f(x)$

governs, for example, Newtonian potential and pressure in incompressible flows.

• Monge-Ampère equation:

$\det D^2 u(x) = f(x, \nabla u(x))$

with potential $u$ and $f$ possibly dependent on position and gradient.

The Monge-Ampère equation is fundamentally nonlinear and degenerate, with central roles in optimal transport (where $f$ encodes mass rearrangement), geometric optics, and geometry. In several recent advances, the analysis and computation of its solutions is recast in terms of sequentially solving linear Poisson-type equations, thereby coupling the two PDEs in an iterative loop (Köhle et al., 31 Aug 2025).

2. Poisson/Monge-Ampère Iterative Algorithms: The L-scheme

A prime example of computational Poisson/Monge-Ampère coupling is the so-called "L-scheme," a robust fixed-point iteration for the elliptic Monge-Ampère equation. For a bounded domain $\Omega\subset \mathbb{R}^d$ with Dirichlet boundary data $\gamma$, the process is as follows:

1. Define Monge-Ampère residual:

$\rho(u)(x) := \det(D^2u(x)) - f(x, \nabla u(x))$

where $u$ is the iterate.

2. Linearize and update: Set $u^{k+1} = u^k + v$, where $v$ solves:

$$\Lambda\,\Delta v = -\rho(u^k) \quad \text{in } \Omega, \qquad v = \gamma - u^k \text{ on } \partial\Omega$$

with $\Lambda$ a scalar parameter typically chosen to majorize the largest eigenvalue of the Hessian to enforce contraction:

$$\Lambda \geq \|\lambda_{\max}(D^2 u^k)\|_{L^\infty}^{d-1}$$

In expanded form:

$$-\Delta u^{k+1} + \Lambda u^{k+1} = \Lambda u^k - \rho(u^k)$$

Each step thus couples the nonlinear operator to the linear Poisson equation with a weighted right-hand side (Köhle et al., 31 Aug 2025).

1. Convergence and Robustness:
   • For classical solutions: linear convergence in the $H^2$-norm is proven, with contraction constant depending on the ratio between the minimal and chosen $\Lambda$.
   • For generalized (viscosity or Alexandrov) solutions: linear convergence in $L^\infty$ norm holds by mollification and domain restriction.
   • The scheme is robust to discretization, nonlinear degeneracies, and initial guess (even non-convex).
2. Acceleration strategies:
   • Precomputed Green’s functions for coarse grids.
   • Preconditioned conjugate gradient (often with algebraic multigrid, AMG).
   • Matrix-free application for large problems.
   • Numerical tests show "mesh-independence" of the number of fixed-point iterations and superior stability and speed over Newton’s method, especially for degenerate or high-frequency cases.

3. Optimal Transport, Gravity, and the Reduction to Poisson

A key theoretical domain where Poisson and Monge-Ampère are coupled is in the theory of optimal transport and its emergence in physical models:

• Monge-Ampère Gravity (MAG):

In cosmological structure formation, the Monge-Ampère equation,

$\det D^2\Phi(x) = \frac{\rho(x)}{\bar{\rho}}$

serves as a nonlinear alternative to Poisson's equation,

$\nabla^2\phi(x) = 4\pi G \rho(x)$

Here, $\Phi$ is the Monge-Ampère potential and $\phi$ the Newtonian potential. The connection is further specified by:

$$\Phi(x) = \frac{\phi(x)}{4\pi G \bar{\rho}} + \frac{|x|^2}{2}$$

In the limit of small perturbations around uniform density, linearizing $\Phi(x)= \frac{1}{2}|x|^2 + \varepsilon \psi(x)$ reduces the Monge-Ampère equation to the Poisson equation for $\psi$, revealing the precise sense in which Monge-Ampère "contains" Poisson as a linearization (Lévy et al., 2024).

• Large deviation and optimal transport derivation:

The continuum Monge-Ampère arises as the deterministic limit of mass-matching $M\gg1$ Brownian particles between initial and final clouds via the large deviation principle. This produces a PDE mass conservation constraint for the transport map, again realized as a Monge-Ampère equation.

4. Quantitative Linearization and Error Analysis

The relationship between Monge-Ampère and Poisson equations is quantified by results that measure the deviation between nonlinear optimal transport and its linearized Poisson analogue:

• When a measure $\mu$ is close to Lebesgue in Wasserstein distance on all scales, the displacement map from optimal coupling is quantitatively approximated by the gradient of the solution to the Poisson equation with source $\Delta u = \mu-1$, with explicit $L^2$ and $L^\infty$ error bounds depending on scale and the Wasserstein deficit. Harmonic approximation lemmas and a Campanato scaling iteration propagate this estimate through scales, allowing sharp justification of formal Monge-Ampère linearization (Goldman et al., 2019).
• Key metrics include
  • Weak norm: optimal plan deviations average $O(\beta(R)/R)$.
  • Strong norm: pointwise error $O(R(\beta(R)/R^2)^{1/(d+2)})$.

This establishes a precise bridge between nonlinear Monge-Ampère dynamics and the linear Poisson regime.

5. Geometric Structures in Fluid Mechanics

In fluid dynamics, specifically incompressible Navier–Stokes, the Poisson equation for pressure emerges naturally; in two dimensions, this can be recast in Monge-Ampère terms:

• On $(M,g)$, the pressure $p$ satisfies (after divergence)

$\Delta_B p = \zeta_{ij}\zeta^{ij} - S_{ij}S^{ij}$

for vorticity $\zeta_{ij}$ and strain $S_{ij}$. In 2D, with stream function $\psi$:

$\nabla^2 p = 2 \det D^2 \psi$

so the Laplacian of pressure is twice the Monge-Ampère operator applied to the stream function.

• Monge-Ampère geometry on the cotangent bundle equips the phase space with a "higher" Lagrangian structure, whose induced metric signature captures the local dominance of vorticity vs. strain. Scalar curvature of the pull-back metric encodes physical phenomena such as vortex accumulation and the topological structure of vortex patches, with Gauss–Bonnet theorem linking curvature integrals to Euler characteristic (Napper et al., 2023).
• Symmetry reductions in 3D (e.g., Arnold-Beltrami-Childress flows, Hill's spherical vortex) relate 3D fluid solutions to Monge-Ampère structure on reduced spaces, aligning geometric, topological, and analytic understanding.

6. Numerical Realizations and Physical Consequences

• Monge-Ampère Gravity:
  • Employs a semi-discrete optimal transport solver using Laguerre cells ("power diagrams") and a damped Newton (discrete Poisson) iteration, accelerated by AMG. This method attains $O(N \log N)$ or better complexity, enabling N-body simulations with over $10^8$ particles.
  • MAG dynamics preferentially stabilize filaments and sheets (due to invariance under volume-preserving affine shears), feature small-scale self-screening (suppression of small halos), but match $\Lambda$CDM at large scales (Lévy et al., 2024).
• L-scheme for Monge-Ampère:
  • Outperforms Newton’s method in regime coverage and mesh-independence; converges even from poor or non-convex initial guesses.
  • AMG-preconditioned finite-difference solvers deliver lowest CPU times and memory footprint; Newton’s method diverges/stalls for fine meshes or poor initializations (Köhle et al., 31 Aug 2025).

7. Summary Table: Poisson/Monge-Ampère Coupling in Practice

Application Area	Poisson PDE Role	Monge-Ampère Coupling Mechanism
L-scheme numerics (Köhle et al., 31 Aug 2025)	Linear subproblem, fixed-point iteration	Nonlinear residual linearized, Poisson step per iteration
Optimal transport (Lévy et al., 2024, Goldman et al., 2019)	Linearization/approximation	Transport map = Monge-Ampère potential, linearized to Poisson
Fluid mechanics (Napper et al., 2023)	Pressure equation	2D pressure as Monge-Ampère of stream function; geometric structure

8. Outlook and Open Directions

Poisson/Monge-Ampère coupling, both as a computational paradigm and as a structural link between linear and nonlinear elliptic PDEs, has prompted advances in robust numerical algorithms, optimal transport theory, geometric fluid mechanics, and cosmological modeling. Robustness of the L-scheme, scaling behavior, and deep links between geometry and physics position this coupling as a foundational element in both analysis and simulation, with ongoing developments in high-dimensional regularity, geometric invariants of turbulence, and nonlinear gravity scenarios.