[Paper Review] Local properties of solutions to non-autonomous parabolic PDEs with state-dependent delays

Alexander V. Rezounenko|arXiv (Cornell University)|Apr 6, 2011

Stability and Controllability of Differential Equations18 references3 citations

TL;DR

This paper establishes local existence, uniqueness, and invariance principles for non-autonomous parabolic PDEs with simultaneous state-dependent delays (both discrete and distributed via Stieltjes integrals). The key contribution is overcoming the non-Lipschitz challenge of state-dependent delays by introducing a novel framework that allows dynamic switching between delay types, enabling rigorous analysis of flexible models in mathematical biology and physics.

ABSTRACT

A wide class of non-autonomous nonlinear parabolic partial differential equations with delay is studied. We allow in our investigations different types of delays such as constant, time-dependent, state-dependent (both discrete and distributed) to be presented simultaneously. The main difficulties arise due to the presence of discrete state-dependent delays since the nonlinear delay term is not Lipschitz on the space of continuous functions. We find conditions for the local existence, uniqueness and study the invariance principle.

Motivation & Objective

• To address the challenge of non-Lipschitz nonlinearities arising from discrete state-dependent delays (SDDs) in non-autonomous parabolic PDEs.
• To develop a unified framework for PDEs with simultaneous discrete and distributed SDDs, allowing dynamic changes in delay type and values along solutions.
• To extend the invariance principle to PDEs with SDDs, particularly for positivity-preserving dynamics in biological and physical models.
• To provide a theoretical foundation for studying asymptotic and qualitative properties of solutions in systems with complex, flexible delay structures.

Proposed method

• Formulates a non-autonomous parabolic PDE with a delay term $ B(t, u_t) = G(t, u(0), F(t, u_t)) $, where $ F(t, ho) $ is a Stieltjes integral combining discrete and distributed SDDs.
• Imposes structural assumptions (A1)–(A5) on the delay kernel $ g $, including state-dependence and regularity, to ensure well-posedness despite non-Lipschitz behavior.
• Applies semigroup theory via a $ C_0 $-semigroup $ \{T(t)\} $ generated by $ -A $, with $ \|T(t)\| \leq e^{\omega t} $, to define mild solutions in a Banach space $ X $.
• Uses a fixed-point argument in the space of continuous functions on $[a-r, b]$, leveraging Gronwall's inequality and uniform Cauchy sequences to prove existence and uniqueness.
• Establishes invariance of closed convex sets $ K \subset X $ (e.g., $ K = [0,\infty)^m $) under the dynamics by verifying the tangency condition: $ \lim_{h \to 0^+} \frac{1}{h} d(\varphi(0) + hB(t,\varphi); K) = 0 $.
• Applies the Lebesgue-Fatou lemma and continuity of $ F_d $ to control convergence of delay terms in the iterative scheme, ensuring uniform convergence of approximating sequences.

Experimental results

Research questions

• RQ1 How can local existence and uniqueness be established for non-autonomous parabolic PDEs with simultaneous discrete and distributed state-dependent delays?
• RQ2 What conditions on the delay kernel $ g $ and nonlinearity $ G $ ensure that the solution remains in a given convex, invariant set (e.g., non-negative solutions)?
• RQ3 How can the invariance principle be extended to PDEs with non-Lipschitz delay terms arising from state-dependent discrete delays?
• RQ4 In what way does the state-dependent nature of both delay values and types (discrete vs. distributed) affect the solution's regularity and continuation properties?
• RQ5 What is the role of the Stieltjes integral formulation in unifying discrete and distributed SDDs and enabling dynamic switching between delay types?

Key findings

• The paper proves the local existence and uniqueness of mild solutions for non-autonomous parabolic PDEs with mixed state-dependent delays, even when the delay term is non-Lipschitz on $ C $.
• A novel invariance principle is established: if the semigroup preserves a closed convex set $ K $ and the vector field satisfies a tangency condition, then solutions remain in $ K $, ensuring nonnegativity in biological models.
• The solution's existence is proven on a maximal interval $[a, b)$ via standard continuation arguments, with uniform convergence of approximating sequences $ \{w^n\} $ on $[a-r, \sigma] $.
• The proof relies on showing that the sequence $ \{v^n\} $ defined by the difference of approximations is uniformly Cauchy, using Gronwall's inequality and the Lebesgue-Fatou lemma to control the delay term.
• The key technical insight is that under assumption (A5), the delay functional $ F_d $ depends only on the initial data $ \varphi $ for $ t \in [a, \sigma] $, enabling uniform convergence of $ F_d(\gamma^n(s), w^{n}_{\gamma^n(s)}) $.
• The result extends the classical invariance theory to PDEs with SDDs, providing a foundation for studying asymptotic behavior and stability in systems with flexible, state-dependent dynamics.