Abstract:This work develops a particle system addressing the approximation of McKean-Vlasov stochastic differential equations (SDEs). The novelty of the approach lies in involving low discrepancy sequences nontrivially in the construction of a particle system with coupled noise and initial conditions. Weak convergence for SDEs with additive noise is proven. A numerical study demonstrates that the novel approach presented here doubles the respective convergence rates for weak and strong approximation of the mean-field limit, compared with the standard particle system. These rates are proven in the simplified setting of a mean-field ordinary differential equation in terms of appropriate bounds involving the star discrepancy for low discrepancy sequences with a group structure, such as Rank-1 lattice points. This construction nontrivially provides an antithetic multilevel quasi-Monte Carlo estimator. An asymptotic error analysis reveals that the proposed approach outperforms methods based on the classic particle system with independent initial conditions and noise.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is to improve the approximation methods of McKean - Vlasov stochastic differential equations (SDEs) by using low - discrepancy sequences, especially to increase the weak and strong convergence rates. Specifically, the author proposes a new method for constructing particle systems, which non - trivially introduces low - discrepancy sequences when constructing particle systems with coupled noise and initial conditions. This method not only improves the convergence speed but also provides an antithetic multilevel quasi - Monte Carlo (MLQMC) estimator.
### Main contributions of the paper:
1. **New particle system construction**: The author proposes a novel method for constructing particle systems, which uses low - discrepancy sequences to generate the initial conditions and noise of the particle systems. This is different from the traditional independent and identically distributed (i.i.d.) method and can significantly improve the convergence rate.
2. **Weak convergence proof**: The author proves that under the condition of additive noise, the weak convergence rate of the new method is twice that of the traditional method. Specifically, for weak and strong approximations, the convergence rates of the new method are \(O(P^{-2})\) respectively, while the convergence rate of the traditional method is \(O(P^{-1})\).
3. **Antithetic multilevel quasi - Monte Carlo (MLQMC)**: The author further extends this method and proposes an antithetic multilevel quasi - Monte Carlo estimator, which performs well in dealing with high - dimensional problems.
4. **Numerical results**: The effectiveness of the new method is verified through numerical experiments, especially in the applications of Ornstein - Uhlenbeck - type equations and Kuramoto oscillators.
### Technical details of the paper:
- **Low - discrepancy sequences**: Low - discrepancy sequences (such as Rank - 1 lattice points) are used to generate the initial conditions and noise of the particle systems. These sequences have good uniform distribution characteristics and help to reduce errors.
- **Wong - Zakai approximation**: By using the Wong - Zakai approximation, the Brownian motion paths are mapped to finite - dimensional points, thus simplifying the proof of weak convergence.
- **Antithetic multilevel method**: Through random shift and stratification techniques, Rank - 1 lattice points are projected onto smaller Rank - 1 lattice point sets, thereby constructing multilevel estimators without introducing additional bias.
### Main conclusions:
- **Convergence rate improvement**: The new method significantly improves the convergence rates in both weak and strong approximations, from \(O(P^{-1})\) to \(O(P^{-2})\).
- **Theoretical and numerical verification**: Through strict theoretical analysis and numerical experiments, the effectiveness and superiority of the new method are verified.
In conclusion, this paper significantly improves the approximation accuracy and convergence speed of McKean - Vlasov SDEs by introducing low - discrepancy sequences and antithetic multilevel quasi - Monte Carlo methods, providing new tools and methods for research in related fields.