A Numerical Algorithm for Solving the Cauchy Problem for a Two-Dimensional Fully Nonlinear Parabolic Equation
Автор: A.A. Chubatov
Журнал: Известия Самарского научного центра Российской академии наук @izvestiya-ssc
Рубрика: Информатика, вычислительная техника и управление
Статья в выпуске: 5 т.27, 2025 года.
Бесплатный доступ
The article continues the research presented in[1]. We solve the Cauchy problem for a two-dimensional fully nonlinear equation arising in the optimization of portfolio investments in the Heston market. First, the fully nonlinear equation is included in a system of quasilinear parabolic equations. The solution of the Cauchy problem for this system is reduced to a stochastic problem in terms of forward-backward stochastic differential equations (FBSDE). Next, the FBSDE is transformed into a stochastic optimal control problem. Finally, this problem is solved using a neural network approach. The proposed numerical algorithm was validated on an example for which it is possible to obtain a quasi-explicit solution.
Fully nonlinear parabolic equation, Cauchy problem, system of quasilinear parabolic equations, forward-backward stochastic differential equations (FBSDE), stochastic optimal control problem, deep learning, forward-backward stochastic neural network (FBSNN)
Короткий адрес: https://sciup.org/148332423
IDR: 148332423 | УДК: 519.633.2 | DOI: 10.37313/1990-5378-2025-27-5-230-239