<p>This paper introduces a meshless barycentric rational interpolation collocation method, in combination with a classical difference formula, for solving optimal control problems governed by nonlinear parabolic equations. A coupled nonlinear algebraic system is derived from the problem using Lagrangian multipliers. The complexity of the system arises from time-fractional derivatives with intrinsically nonlocal nature and historical dependence, and mutually coupled state and co-state variables evolving in opposite temporal directions. The temporal discretization is based on a classical L1 approximation for the Caputo derivative, together with Newton linearization for the nonlinear term. Error estimates for the linearized time-discrete schemes of the state and co-state equations are presented. With the aim of enhancing spatial accuracy, a high-precision meshless barycentric rational collocation method is considered as the spatial discretization technique. Furthermore, we present comprehensive consistency analyses of both spatially discrete and fully discrete schemes, utilizing the approximation properties of the collocation method. Finally, several numerical examples, encompassing one, two, and three dimensions, with power-law or exponential-type nonlinear terms, are provided to demonstrate the effectiveness and reliability of the collocation method on uniform and nonuniform nodes.</p>

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Numerical approximation of nonlinear time fractional parabolic optimal control problems based on a collocation technique

  • Rong Huang,
  • Zhifeng Weng,
  • Jianhua Yuan

摘要

This paper introduces a meshless barycentric rational interpolation collocation method, in combination with a classical difference formula, for solving optimal control problems governed by nonlinear parabolic equations. A coupled nonlinear algebraic system is derived from the problem using Lagrangian multipliers. The complexity of the system arises from time-fractional derivatives with intrinsically nonlocal nature and historical dependence, and mutually coupled state and co-state variables evolving in opposite temporal directions. The temporal discretization is based on a classical L1 approximation for the Caputo derivative, together with Newton linearization for the nonlinear term. Error estimates for the linearized time-discrete schemes of the state and co-state equations are presented. With the aim of enhancing spatial accuracy, a high-precision meshless barycentric rational collocation method is considered as the spatial discretization technique. Furthermore, we present comprehensive consistency analyses of both spatially discrete and fully discrete schemes, utilizing the approximation properties of the collocation method. Finally, several numerical examples, encompassing one, two, and three dimensions, with power-law or exponential-type nonlinear terms, are provided to demonstrate the effectiveness and reliability of the collocation method on uniform and nonuniform nodes.