A probabilistic interpretation of a non-conservative and path-dependent nonlinear reaction-advection-diffusion system

Basic Information

• Paper ID: 2407.19301
• Title: A probabilistic interpretation of a non-conservative and path-dependent nonlinear reaction-advection-diffusion system
• Authors: Daniela Morale, Leonardo Tarquini, Stefania Ugolini
• Classification: math.PR (Probability Theory)
• Publication Date: October 14, 2025
• Paper Link: https://arxiv.org/abs/2407.19301

Abstract

This paper investigates a reaction-advection-diffusion system describing sulfation phenomena, deriving a single regularized non-conservative and path-dependent nonlinear partial differential equation. A probabilistic interpretation is proposed through non-Markovian McKean-Vlasov stochastic differential equations coupled with Feynman-Kac type equations. The paper discusses the well-posedness of the stochastic model and establishes propagation of chaos properties for the associated interacting particle system.

Research Background and Motivation

Problem Background

This research originates from the marble degradation problem in cultural heritage preservation. Specifically, when chemical substances in polluted air (particularly sulfur dioxide SO₂) diffuse through stone pores and interact with pore surfaces, chemical reactions produce gypsum, a more porous material than marble, thereby causing material degradation.

Mathematical Model

The original PDE-ODE system describing sulfation phenomena is:

∂ₜρ = ∇ · (φ(c)∇s) - λρc
∂ₜc = -λρc

where:

• ρ: sulfur dioxide concentration
• c: calcite density
• φ(c) = φ₀ + φ₁c: material porosity
• λ: reaction rate

Research Motivation

1. Theoretical Challenges: Existing models possess non-conservative and path-dependent properties, lacking deep probabilistic theoretical interpretation
2. Multi-scale Understanding: Establishing connections between macroscopic PDE models and microscopic particle dynamics
3. Numerical Computation: Probabilistic representation provides new computational frameworks for numerical methods

Core Contributions

1. Establishing Probabilistic Representation: Providing complete probabilistic interpretation for non-conservative and path-dependent reaction-diffusion equations
2. Theoretical Analysis: Proving well-posedness of McKean-Feynman-Kac stochastic differential equations
3. Regularization Methods: Introducing regularization techniques with non-local dependence to handle singularities
4. Propagation of Chaos: Establishing and proving propagation of chaos properties for interacting particle systems
5. Multi-scale Connection: Constructing rigorous mathematical bridges between microscopic particle dynamics and macroscopic PDEs

Methodology Details

Task Definition

Converting the original PDE-ODE system into a single path-dependent PDE and establishing its probabilistic representation. Specifically, finding a stochastic process Y such that the evolution of its distribution follows the given PDE.

Model Architecture

1. PDE Reconstruction

First rewriting the original system as a single equation for ρ:

∂ₜρ(t,x) = Δρ(t,x) - ∇·(b(ρ̄(·,x)(t), ∇ρ̄(·,x)(t))ρ(t,x)) - λc₀exp(-λρ̄(·,x)(t))ρ(t,x)

where ρ̄(·,x)(t) = ∫₀ᵗ ρ(s,x)ds is the integral functional.

2. Regularization Treatment

To handle path-dependence, introducing regularization with convolution kernel K:

∂ₜρ(t,x) = Δρ(t,x) - ∇·[b(K*ρ̄(·,x)(t), ∇K*ρ̄(·,x)(t))ρ] - λc₀exp(-λK*ρ̄(·,x)(t))ρ(t,x)

3. McKean-Feynman-Kac SDE

Constructing the stochastic differential equation:

Yₜ = Y₀ + ∫₀ᵗ b(uᵐ(·,Yₛ)(s), ∇uᵐ(·,Yₛ)(s))ds + √2Wₜ

where uᵐ satisfies the Feynman-Kac type equation:

uᵐ(t,y) = E[K(y-Yₜ)exp(-λc₀∫₀ᵗ exp(-λ∫₀ˢ uᵐ(r,Yₛ)dr)ds)]

Technical Innovations

1. Non-Markovian Treatment

• Introducing integral functional ρ̄(·,x)(t) to capture path-dependence
• Designing special Banach space norm ‖·‖∞,1,1 considering discount terms and integral operators

2. Regularization Techniques

• Using smooth kernel K to handle singularities
• Proving recovery of the original equation when K is the Dirac delta function

3. Existence and Uniqueness Proof

• Utilizing contraction mapping theorem in appropriate Banach spaces to prove solution existence and uniqueness
• Establishing continuity of solutions with respect to measures and spatial variables

Experimental Setup

Theoretical Verification Framework

This work is primarily theoretical, with verification through:

1. Well-posedness Analysis: Proving existence, uniqueness, and regularity of solutions to Feynman-Kac equations
2. Stability Estimates: Establishing continuity estimates of solutions with respect to measure perturbations
3. Particle Approximation: Constructing N-particle systems and proving their convergence

Mathematical Tools

• Wasserstein distance measuring differences between probability measures
• Grönwall's lemma handling integral inequalities
• Fixed point theorems proving existence

Experimental Results

Main Theoretical Results

1. Well-posedness of Feynman-Kac Equation (Proposition 2.7)

For any m ∈ P(C), equation (23) admits a unique solution uᵐ ∈ C₊.

2. Regularity Estimates (Proposition 2.8)

The solution uᵐ satisfies:

• Boundedness: |uᵐ| ≤ Mₖ
• Lipschitz continuity: |uᵐ(t,y) - uᵐ(t,y')| ≤ Lₖ|y-y'|

3. Stability Estimates (Proposition 2.11)

|uᵐ(t,y) - uᵐ'(t,y')|² ≤ C₁(t)(|y-y'|² + D²ₜ(m,m'))

4. Well-posedness of McKean-Vlasov SDE (Proposition 3.3)

SDE (9) admits a pathwise unique strong solution, and weak solutions are unique in the sense of probability law.

5. Propagation of Chaos (Theorem 4.5)

The N-particle system converges to the independent particle system, with propagation of chaos property holding.

Convergence Rate

The convergence rate of the particle system is O(1/N):

E[‖uᵘᴺ - uᵐ‖²∞] + sup E[sup |ξⁱₛ - Yⁱₛ|²] ≤ C/N

Related Work

Theoretical Foundations

1. McKean-Vlasov Theory: Classical nonlinear SDE theoretical framework
2. Feynman-Kac Formula: Probabilistic representation theory between PDEs and SDEs
3. Particle Systems: Large deviation principles and propagation of chaos theory

Application Domains

1. Chemotaxis Models: Probabilistic interpretation of Keller-Segel equations
2. Cultural Heritage Preservation: Mathematical modeling of stone degradation
3. Reaction-Diffusion Systems: Stochastic representation of non-conservative PDEs

Technical Development

This work extends the work of Le Cavil et al., addressing more complex path-dependent cases.

Conclusions and Discussion

Main Conclusions

1. Successfully establishing complete probabilistic representation for non-conservative path-dependent PDEs
2. Proving mathematical well-posedness of the corresponding stochastic model
3. Establishing rigorous connections between microscopic and macroscopic scales

Limitations

1. Regularization Dependence: Requiring smooth kernel K; direct treatment of the original equation remains challenging
2. One-dimensional Restriction: Current results limited to one-dimensional cases
3. Initial Value Conditions: Requiring specific regularity of initial values

Future Directions

1. Extending to multi-dimensional cases
2. Investigating convergence of numerical algorithms
3. Exploring more general non-conservative systems

In-depth Evaluation

Strengths

1. Theoretical Rigor: Providing complete mathematical proof framework
2. Methodological Innovation: Connecting path-dependent PDEs with non-Markovian SDEs
3. Applied Value: Providing new mathematical tools for cultural heritage preservation
4. Technical Depth: Addressing challenging non-conservative and path-dependent properties

Weaknesses

1. Computational Complexity: Practical numerical implementation may face challenges
2. Assumption Conditions: Requiring relatively strong regularity assumptions
3. Application Verification: Lacking validation with real data

Impact

1. Theoretical Contribution: Providing new perspectives for non-conservative PDE theory
2. Interdisciplinary Value: Connecting probability theory, PDE theory, and applied mathematics
3. Computational Potential: Providing theoretical foundation for Monte Carlo methods

Applicable Scenarios

1. Degradation modeling in cultural heritage preservation
2. Chemical reaction-diffusion processes
3. Stochastic modeling of non-conservative systems

References

The paper cites 32 relevant references, covering multiple fields including McKean-Vlasov theory, Feynman-Kac formula, and mathematical modeling of cultural heritage. Particularly noteworthy are the works by Le Cavil et al. on non-conservative McKean-type equations 11,15,18 and research by Talay and Tomašević on Keller-Segel models 25.