Shirui Zheng1, Xin Gui2
1Zhizhen College, Beihang University, Beijing, 100191, China
2School of Civil Engineering, Shandong Jianzhu University, Jinan, 250101, China
Inspired by the classification idea of the K-nearest neighbors (KNN) algorithm, this paper proposes a new method for numerically solving partial differential equations (PDE) using the Feynman-Kac formula. The method involves establishing a connection between PDE and stochastic differential equations through the Feynman-Kac formula. Random points are selected within the domain of the equation, and the KNN algorithm is used to partition these points into different regions. Then, Monte Carlo simulation is performed on the random points within each region to obtain a series of simulated values. These simulated values are substituted into the corresponding Feynman-Kac formula for each random point to obtain the solution. Finally, within each region, the average of the solutions for all random points belonging to that region is calculated, resulting in the corresponding approximate solution. By selecting an appropriate partitioning approach, a higher-precision global solution to the PDE can be constructed. Through a series of numerical simulations, the results show that the PDE global solution constructed by the new method achieves higher accuracy compared to traditional interpolation methods.
Monte Carlo Simulation, Feynman-Kac formula, Numerical method of PDE, KNN
Shirui Zheng, Xin Gui. A modified method based on the K-nearest neighbor approach for solving PDE global solutions using the Feynman-Kac formula. Academic Journal of Mathematical Sciences (2023) Vol. 4, Issue 4: 36-44. https://doi.org/10.25236/AJMS.2023.040406.
 Jakubowski J, Wiśniewolski M. On matching diffusions, Laplace transforms and partial differential equations [J]. Stochastic Processes and their Applications, 2015, 125(10): 3663-3690.
 Gavete L, Ureña F, Benito J J, et al. Solving second order non-linear elliptic partial differential equations using generalized finite difference method [J]. Journal of Computational and Applied Mathematics, 2017, 318: 378-387.
 Ma C, Zheng W. A fourth-order unfitted characteristic finite element method for free-boundary problems [J]. Journal of Computational Physics, 2022, 469: 111552.
 Zhang S. Cost-sensitive KNN classification [J]. Neurocomputing, 2020, 391: 234-242.
 Jasra A, Jo S, Nott D, et al. Multilevel Monte Carlo in approximate Bayesian computation[J]. Stochastic Analysis and Applications, 2019, 37(3): 346-360.
 Ferré G, Stoltz G. Error estimates on ergodic properties of discretized Feynman–Kac semigroups [J]. Numerische Mathematik, 2019, 143: 261-313.
 Hutzenthaler M, Jentzen A, Kruse T. On multilevel Picard numerical approximations for high-dimensional nonlinear parabolic partial differential equations and high-dimensional nonlinear backward stochastic differential equations [J]. Journal of Scientific Computing, 2019, 79(3): 1534-1571.
 Ma T, Cao C. L1 adaptive control for general partial differential equation (PDE) systems [J]. International Journal of General Systems, 2019, 48(6): 656-689.
 Li J, Cheng Y. Barycentric rational interpolation method for solving KPP equation [J]. Electronic Research Archive, 2023, 31(5): 3014-3029.
 Arendt W, Urban K. Partial Differential Equations: An Introduction to Analytical and Numerical Methods [M]. Springer Nature, 2023.
 Xing W, Bei Y. Medical health big data classification based on KNN classification algorithm [J]. IEEE Access, 2019, 8: 28808-28819.
 Scorolli R. Feynman-Kac formula for the heat equation driven by time-homogeneous white noise potential [J]. arXiv preprint arXiv: 2108.12406, 2021.