Abstract:We study the propagation of chaos in a class of moderately interacting particle systems for the approximation of singular kinetic McKean-Vlasov SDEs driven by alpha-stable processes. Diffusion parts include Brownian (alpha=2) and pure-jump (1<\alpha<2) perturbations and interaction kernels are considered in a non-smooth anisotropic Besov space. Using Duhamel formula, sharp density estimates (recently issued in Hao, Rockner and Zhang 2023), and suitable martingale functional inequalities, we obtain direct estimates on the convergence rate between the empirical measure of the particle systems toward the McKean-Vlasov distribution. These estimates further lead to quantitative propagation of chaos results in the weak and strong sense.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is about the chaos propagation problem of moderately interacting particle systems in singular kinetic McKean - Vlasov stochastic differential equations (McKean - Vlasov SDEs). Specifically, the researchers are concerned with how to approximate the singular kinetic McKean - Vlasov stochastic differential equations driven by α - stable processes through moderately interacting particle systems, and quantify the chaos propagation properties of this approximation in the weak sense and the path - wise sense.
### Core Problems of the Paper
1. **Quantifying Chaos Propagation**:
- The researchers aim to establish the chaos propagation properties between moderately interacting particle systems and singular kinetic McKean - Vlasov stochastic differential equations, and quantify these properties.
- Specifically, they hope to estimate the rate at which the empirical measure of the particle system converges to the McKean - Vlasov distribution.
2. **Moderately Interacting Particle Systems**:
- To deal with the definitional difficulties brought by the singular kernel, the researchers introduce moderately interacting particle systems, in which the initial interaction kernel is replaced by its convolution with an appropriate smoothing function.
- This method makes the particle system itself well - defined, so that the analysis of chaos propagation properties can be carried out.
3. **Singular Kinetic McKean - Vlasov Equations**:
- This equation describes the evolution of the momentum of an object in a possibly abnormal medium under the action of a distribution - dependent force field.
- The form of the equation is:
\[
\ddot{X}_t=(b_t * \mu_t)(X_t,\dot{X}_t)+\dot{L}^\alpha_t,\quad(X_0,\dot{X}_0)\sim\mu_0,\quad0\leq t\leq T
\]
where \(\mu_t\) is the joint distribution of \((X_t,\dot{X}_t)\), and \(L^\alpha\) is a \(d\)-dimensional isotropic α - stable Lévy process.
### Main Results
- **Weak Chaos Propagation**:
- Under appropriate assumptions, the researchers prove that the total variation distance between the empirical measure of the particle system and the McKean - Vlasov distribution converges at a certain rate.
- The specific convergence rate depends on the parameters \(\beta\) and \(\zeta\), and the value ranges of these parameters are restricted by some technical conditions.
- **Strong Chaos Propagation**:
- If it is further assumed that \(b\) belongs to the mixed Besov space \(L^\infty([0, T_0], B^{s_0, s_1}_{p_0; x, a})\) and satisfies certain regularity conditions, then the path - wise chaos propagation property can be obtained.
- The convergence rate in the path - wise sense is also given in a certain form.
### Technical Challenges
- **Handling of Singular Kernels**:
- The existence of singular kernels makes it difficult to directly define the particle system, so moderately interacting particle systems are introduced.
- Through the convolution operation, the definitional problems brought by singular kernels can be avoided, so that the particle system itself is well - defined.
- **Quantification of Chaos Propagation**:
- Quantifying the chaos propagation properties requires an accurate estimate of the difference between the empirical measure of the particle system and the McKean - Vlasov distribution.
- This involves the a priori regularity control of the solution of the Fokker - Planck equation, and the martingale - type inequalities in an appropriate Banach space.
### Conclusion
By introducing moderately interacting particle systems, this paper successfully solves the chaos propagation problem of singular kinetic McKean - Vlasov stochastic differential equations and quantifies this chaos propagation property. These results are not only of great theoretical significance, but also provide a theoretical basis for numerical simulation and practical applications.