On some nonlocal, nonlinear diffusion problems

This note is devoted to some nonlocal, nonlinear elliptic problems with an emphasis on the computation of the solution of such problems, reducing it in particular to a fixed point argument in R. Errors estimates and numerical experiments are provided.

💡 Research Summary

**
The paper investigates a class of stationary, non‑local, nonlinear diffusion problems of elliptic type. Starting from the classical reaction‑diffusion equation
∂ₜu – a Δu + λ u = F,
the authors replace the constant diffusion coefficient a and the death rate λ by functions that may depend on space, on the local value of the unknown, and, crucially, on a global functional of the solution. The most general non‑local model considered is
a(x, ∫Ω u) Δu + λ(x, ∫Ω u) u = F,
which captures the idea that the whole population influences the mobility of individuals.

Setting the time derivative to zero leads to the elliptic problem
–a(x, u) Δu + λ(x, u) u = F in Ω, u = 0 on ∂Ω.
The authors introduce a scalar functional ℓ : H₀¹(Ω) → ℝ (for example, an Lᵖ‑norm or an integral of u) and rewrite the non‑local dependence as a parameter μ = ℓ(u). For each fixed μ, the problem becomes a linear Poisson‑type equation with coefficients A(·, μ) and λ(·, μ). The key observation (Theorem 2.2) is that solving the original non‑local problem is equivalent to finding a fixed point of the scalar map
μ = ℓ(u_{S,μ}),
where u_{S,μ} denotes the solution of the linearized problem in a closed subspace S ⊂ H₀¹(Ω). Lemma 2.3 proves the continuity of the mapping μ ↦ u_{S,μ} in H₀¹, and Theorem 2.4 guarantees existence of at least one solution provided ℓ is continuous and bounded on bounded sets. In the special case λ = 0 and A(x, r) = a(r) I, with ℓ homogeneous of degree p, the fixed‑point equation reduces to
μ = ℓ(ψ_S) a(μ)ᵖ,
where ψ_S solves the standard Poisson problem. This reduction yields an explicit existence and uniqueness result (Corollary 2.5).

Section 3 extends the analysis to a fully nonlinear setting where the diffusion matrix A(u) and reaction coefficient λ(u) are continuous operators from H₀¹(Ω) into L^∞(Ω). The authors formulate a Galerkin discretization on a finite‑dimensional subspace S and prove existence of a discrete solution via Brouwer’s fixed‑point theorem (Theorem 3.1). Under a Lipschitz condition on A (inequality 3.6) and a small‑data assumption (3.7), uniqueness of the discrete solution follows (Theorem 3.2). The results are then lifted to the infinite‑dimensional problem by a density argument, establishing existence of a weak solution for the original non‑local nonlinear elliptic equation (Theorem 3.4).

Section 4 focuses on the practical computation of the fixed point. The scalar map G(x) := ℓ(u_x) is iterated by the simple scheme x_{n+1}=G(x_n). The authors provide rigorous convergence criteria: if there exist ν₁ < μ₀ < ν₂ such that x < G(x) < μ₀ for x∈(ν₁, μ₀) and μ₀ < G(x) < x for x∈(μ₀, ν₂), then any initial guess x₀∈(ν₁, ν₂) generates a monotone sequence converging to μ₀. They also discuss pathological situations where G possesses multiple fixed points, leading to oscillations or divergence (illustrated in Figures 2 and 3).

For the numerical implementation, the authors propose the following algorithm: (i) solve the linear Poisson problem on the Galerkin space S to obtain ψ_S; (ii) compute ℓ(ψ_S); (iii) update the scalar parameter using the reduced fixed‑point equation μ = ℓ(ψ_S) a(μ)ᵖ; (iv) recover the approximate solution u_n^S = ψ_S / a(μ_n). When a(r) is bounded away from zero and infinity, the scalar fixed‑point equation always admits at least one solution (Remark 4.2).

Section 5 (partially presented) is devoted to error analysis. Building on the continuity and boundedness assumptions, the authors intend to derive a priori error estimates for the Galerkin approximation and to quantify the convergence rate of the fixed‑point iteration. The combination of analytical existence/uniqueness results, a clear reduction to a scalar fixed‑point problem, and a concrete Galerkin‑based numerical scheme makes the paper a valuable contribution to the study of non‑local, nonlinear diffusion phenomena.