Local and Global Existence of A Nonlocal Equation with A Singular Integral Drift Term

Yingdong Lu

TL;DR

This paper establishes the local and global existence of solutions for a nonlocal evolution equation featuring a fractional Laplacian and a singular drift term. By leveraging both fixed-point theory and probabilistic methods (stochastic differential equations), the author extends previous existence results to a broader class of singular integral operators known as Calderón-Zygmund kernels.

What Problem Does This Paper Address?

The paper investigates a specific type of initial value problem: $$\partial_t u(t, x) = -(-\Delta)^{\alpha/2}u - \nabla \cdot (u B(u))$$ This equation describes how a density $u$ evolves over time under the influence of fractional diffusion (the fractional Laplacian) and a drift term $B(u)$.

The challenge lies in the drift term $B(u)$, which is defined by a singular integral operator. While previous research has handled "regular" drift terms (where the kernel is smooth or well-behaved), this paper tackles the case where the drift is a Calderón-Zygmund operator. These operators are "singular," meaning they can blow up as the distance between points approaches zero, making it much harder to prove that a solution exists and remains stable over time. This type of equation is physically relevant in modeling fractal interfaces in statistical mechanics and generalizations of the Kardar–Parisi–Zhang (KPZ) model.

Methods

The author employs a two-pronged mathematical strategy:

  1. Analytic Approach (Local Existence): To prove the solution exists for at least a short amount of time, the author uses the Banach Fixed Point Theorem. They define a bilinear mapping based on the equation's integral form and prove that this mapping is a "contraction" within a specific function space ($L^p_x L^\infty_t$). This requires careful estimation of the singular integral's bounds.
  2. Probabilistic Approach (Global Existence): To prove the solution exists for all time ($T < \infty$), the author shifts to the world of probability. They link the partial differential equation (PDE) to a Stochastic Differential Equation (SDE) driven by an $\alpha$-stable process (a type of jump process). By showing that a unique solution exists for the SDE and that its probability density function satisfies the original PDE, they establish global existence. This involves using the Wasserstein distance to measure the "closeness" of probability measures during an iterative construction process.

Datasets

Not discussed in the paper. As a theoretical mathematics paper, it focuses on proofs and derivations rather than empirical data or benchmarks.

Key Insights and Ideas

  • Bridging Analysis and Probability: The most significant "aha moment" is the use of SDEs to solve a deterministic PDE. By viewing the density $u$ as the law of a stochastic process, the author can use tools from stochastic calculus to bypass some of the difficulties of singular integrals in pure analysis.
  • Handling Singularity: The paper identifies that while Calderón-Zygmund operators are "weaker" in terms of boundedness than regular kernels, they still possess enough structure (specifically their boundedness on $L^p$ and Lipschitz spaces) to allow for existence proofs if the function spaces are chosen correctly.
  • Kernel Constraints: The global existence proof relies on specific conditions for the Calderón-Zygmund kernel: it must have a regularity parameter $\delta > 1$ and satisfy the cancellation property $K_b 1 = 0$.

Key Results

  • Theorem 1 (Local Existence): For any $p > 2$ and initial data $u_0 \in L^p_x$, there exists a time $T^* > 0$ such that a unique weak solution exists on the interval $[0, T^*]$.
  • Theorem 2 (Global Existence): Under stricter conditions—specifically that the kernel parameter $\delta > 1$, the operator annihilates constants ($K_b 1 = 0$), and the initial condition is Lipschitz—the solution exists globally for any time horizon $T < \infty$.
  • Quantitative Estimates: The paper provides explicit bounds (Lemmas 1 and 3) showing how the fractional heat kernel interacts with the singular drift, proving that the nonlinear term doesn't "overpower" the diffusion term too quickly.

Limitations and Future Work

  • Parameter Restrictions: The global existence result requires $\delta > 1$. In many standard Calderón-Zygmund contexts, $\delta$ is only assumed to be $> 0$. Extending this to the full range of $0 < \delta \leq 1$ remains an open challenge.
  • Initial Condition Smoothness: Global existence currently requires the initial condition $u_0$ to be a Lipschitz function. Future work could investigate if this requirement can be relaxed to less smooth spaces (like $L^p$).
  • Numerical Verification: As a purely theoretical work, the paper does not provide numerical simulations to visualize how these singular drifts affect the evolution of the density $u$ in practice.

Summary updated

Abstract

We study an initial value problem with fractional Laplacian and a singular drift term, and obtain local and global existence theorems similar to the results in Jourdain et al.(2005).
BibTeX
@article{lu2021local,
  author = {Yingdong Lu},
  title = {{Local and Global Existence of A Nonlocal Equation with A Singular Integral Drift Term}},
  journal = {The Journal of Nonlinear Sciences and Applications},
  year = {2021},
  doi = {10.22436/jnsa.015.01.05},
}
Figure