[Paper Review] Strong invariance and noise-comparison principles for some parabolic stochastic PDEs

Mathew Joseph, Davar Khoshnevisan+1 more|White Rose Research Online (University of Leeds, The University of Sheffield, University of York)|Apr 28, 2014

Stochastic processes and financial applications7 references4 citations

TL;DR

This paper establishes strong invariance principles and noise-comparison theorems for parabolic stochastic PDEs, particularly the stochastic heat equation with space-time white noise and a fractional Laplacian operator. By approximating the continuum equation via rescaled interacting diffusions on a lattice, the authors prove that product moments of solutions with different nonlinearities satisfy comparison inequalities under appropriate conditions, extending known results to a broader class of stochastic PDEs with stable Lévy noise.

ABSTRACT

We consider a system of interacting diffusions on the integer lattice. By letting the mesh size go to zero and by using a suitable scaling, we show that the system converges (in a strong sense) to a solution of the stochastic heat equation on the real line. As a consequence, we obtain comparison inequalities for product moments of the stochastic heat equation with different nonlinearities.

Motivation & Objective

• To establish comparison inequalities for product moments of solutions to the stochastic heat equation with different nonlinearities.
• To prove that solutions of lattice-based interacting diffusions converge strongly to the solution of the continuum stochastic heat equation as the mesh size tends to zero.
• To extend comparison theorems for SPDEs beyond the standard case, incorporating general Lévy noise via fractional Laplacian operators.
• To provide a rigorous approximation framework for stochastic PDEs using discrete interacting particle systems.
• To derive sharp moment estimates for the stochastic heat equation with linear and nonlinear noise terms.

Proposed method

• Use a rescaled system of interacting diffusions on a lattice with generator derived from a symmetric α-stable Lévy process.
• Apply a space-time scaling limit where the mesh size ε → 0 and the jump rate is speeded up to recover the continuum stochastic heat equation.
• Leverage the strong convergence of the rescaled lattice system to the solution of the stochastic heat equation with space-time white noise.
• Use moment comparison theorems for interacting diffusions (Cox, Fleischmann, Greven) as a key tool in the limit.
• Apply theorems on Hölder continuity and moment bounds for SPDEs (e.g., Dalang, Foondun–Khoshnevisan) to control the convergence rate.
• Use Chebyshev’s inequality and moment estimates to control the probability of large deviations in the approximation error.

Experimental results

Research questions

• RQ1 Under what conditions do solutions to the stochastic heat equation with different nonlinearities satisfy moment comparison inequalities?
• RQ2 Can the solution of a lattice-based interacting diffusion system converge strongly to the solution of the continuum stochastic heat equation with space-time white noise?
• RQ3 How does the choice of the fractional Laplacian operator (with α ∈ (1,2]) affect the regularity and moment behavior of the solution?
• RQ4 What is the rate of convergence of the discrete approximation to the continuum solution in terms of the mesh size ε?
• RQ5 Can sharp lower bounds on the moments of the solution be derived using the comparison principle and known moment estimates?

Key findings

• The solution to the stochastic heat equation with space-time white noise and a fractional Laplacian operator is almost surely Hölder-continuous and admits finite moments of all orders.
• For any t > 0 and spatial points x₁,…,xₘ ∈ ℝ, if σ ≤ σ̄ pointwise, then E[∏uₜ(xᵢ)] ≤ E[∏ūₜ(xᵢ)] holds almost surely.
• The rescaled interacting diffusion system on εℤ converges strongly to the solution of the stochastic heat equation in the limit ε → 0, with error probability o(ε^Q) for any Q > 0.
• The convergence is uniform over compact space-time regions, and the error bound is independent of the initial condition (constant 1).
• A sharp lower bound on the k-th moment of the solution is established: E[|uₜ(x)|^k] ≥ exp(Lσ⁴k(k²−1)t/(48ν)) for linear σ(x) = Lσx.
• The comparison principle for product moments is extended to general Lipschitz nonlinearities via the approximation framework and moment comparison theorems for interacting diffusions.