Table of Contents
Fetching ...

The Porous Medium Equation: Multiscale Integrability in Large Deviations

Benjamin Gess, Daniel Heydecker

TL;DR

The article develops a robust multiscale framework to prove a large deviations principle for a zero-range process with unbounded and degenerate diffusion that converges to the porous medium equation $\partial_t u=\tfrac12\Delta(u^{\alpha})$ with $\alpha>1$. By introducing a scale-separated integrability mechanism, the authors bridge pathwise regularity and regularity of laws using a Feynman–Kac variational approach and Orlicz-norm control, enabling uniform integrability of $u^\alpha$ across all vanishing particle-size regimes $χ_N\to0$. The main technical feat is a multiscale recombination argument that propagates integrability gains from macroscopic to mesoscopic and microscopic scales, complemented by a novel pathwise-regularity lemma and a hierarchy of block-averaging estimates. Exponential tightness and a superexponential estimate are established, yielding the LDP with the same rate function $\mathcal{I}_ρ$ as in prior rescaled results, but now valid in any scaling regime, including regimes with highly degenerate noise. The work thus advances the macroscopic fluctuation theory for degenerate nonlinear diffusions and provides a versatile template for LDPs in SPDEs with unbounded nonlinearities and scale-dependent noise.

Abstract

We consider a zero-range process $η^N_t(x)$ with superlinear local jump rate, which in a hydrodynamic-small particle rescaling converges to the porous medium equation $\partial_t u=\frac12Δu^α, α>1$. As a main result we obtain a large deviation principle in any scaling regime of vanishing particle size $χ_N\to 0$. The key challenge is to develop uniform integrability estimate on the nonlinearity $(η^N(x))^α$ in a situation where neither pathwise regularity nor Dirichlet-form based regularity is readily available. We resolve this by introducing a novel multiscale argument exploiting the appearance of pathwise regularity across scales.

The Porous Medium Equation: Multiscale Integrability in Large Deviations

TL;DR

The article develops a robust multiscale framework to prove a large deviations principle for a zero-range process with unbounded and degenerate diffusion that converges to the porous medium equation with . By introducing a scale-separated integrability mechanism, the authors bridge pathwise regularity and regularity of laws using a Feynman–Kac variational approach and Orlicz-norm control, enabling uniform integrability of across all vanishing particle-size regimes . The main technical feat is a multiscale recombination argument that propagates integrability gains from macroscopic to mesoscopic and microscopic scales, complemented by a novel pathwise-regularity lemma and a hierarchy of block-averaging estimates. Exponential tightness and a superexponential estimate are established, yielding the LDP with the same rate function as in prior rescaled results, but now valid in any scaling regime, including regimes with highly degenerate noise. The work thus advances the macroscopic fluctuation theory for degenerate nonlinear diffusions and provides a versatile template for LDPs in SPDEs with unbounded nonlinearities and scale-dependent noise.

Abstract

We consider a zero-range process with superlinear local jump rate, which in a hydrodynamic-small particle rescaling converges to the porous medium equation . As a main result we obtain a large deviation principle in any scaling regime of vanishing particle size . The key challenge is to develop uniform integrability estimate on the nonlinearity in a situation where neither pathwise regularity nor Dirichlet-form based regularity is readily available. We resolve this by introducing a novel multiscale argument exploiting the appearance of pathwise regularity across scales.
Paper Structure (50 sections, 14 theorems, 150 equations, 1 figure)

This paper contains 50 sections, 14 theorems, 150 equations, 1 figure.

Key Result

Theorem 1

Let $\chi_N \to 0$. Fix $\rho\in C(\mathbb{T}^d, (0,\infty))$, and let $(\Omega, \mathcal{F},\mathbb{P})$ be a probability space on which are defined, for all $N$, a rescaled zero-range process $\eta^N_\bullet$ with parameter $\alpha\ge 1$ and initial data distributed according to the slowly varying

Figures (1)

  • Figure 1: Two steps of the scheme.

Theorems & Definitions (23)

  • Theorem 1
  • Theorem 2
  • Theorem 3
  • Proposition 1.1
  • Proposition 3.1: Integrability via Dirichlet Form
  • Proposition 3.2
  • Lemma 3.3
  • Lemma 3.4
  • Lemma 3.5
  • Lemma 3.6
  • ...and 13 more