This paper proposes a deep unfitted Nitsche method for solving elliptic interface problems with high contrasts in high dimensions.To capture discontinuities of the solution caused by interfaces,we reformulate the prob...This paper proposes a deep unfitted Nitsche method for solving elliptic interface problems with high contrasts in high dimensions.To capture discontinuities of the solution caused by interfaces,we reformulate the problem as an energy minimization problem involving two weakly coupled components.This enables us to train two deep neural networks to represent two components of the solution in highdimensional space.The curse of dimensionality is alleviated by using theMonte-Carlo method to discretize the unfittedNitsche energy functional.We present several numerical examples to show the performance of the proposed method.展开更多
A Nitsche-based element-free Galerkin(EFG)method for solving semilinear elliptic problems is developed and analyzed in this paper.The existence and uniqueness of the weak solution for semilinear elliptic problems are ...A Nitsche-based element-free Galerkin(EFG)method for solving semilinear elliptic problems is developed and analyzed in this paper.The existence and uniqueness of the weak solution for semilinear elliptic problems are proved based on a condition that the nonlinear term is an increasing Lipschitz continuous function of the unknown function.A simple iterative scheme is used to deal with the nonlinear integral term.We proved the existence,uniqueness and convergence of the weak solution sequence for continuous level of the simple iterative scheme.A commonly used assumption for approximate space,sometimes called inverse assumption,is proved.Optimal order error estimates in L 2 and H1 norms are proved for the linear and semilinear elliptic problems.In the actual numerical calculation,the characteristic distance h does not appear explicitly in the parameterβintroduced by the Nitsche method.The theoretical results are confirmed numerically。展开更多
基金supported by Andrew Sisson Fund of the University of MelbourneX.Y.was partially supported by the NSF grants DMS-1818592 and DMS-2109116.
文摘This paper proposes a deep unfitted Nitsche method for solving elliptic interface problems with high contrasts in high dimensions.To capture discontinuities of the solution caused by interfaces,we reformulate the problem as an energy minimization problem involving two weakly coupled components.This enables us to train two deep neural networks to represent two components of the solution in highdimensional space.The curse of dimensionality is alleviated by using theMonte-Carlo method to discretize the unfittedNitsche energy functional.We present several numerical examples to show the performance of the proposed method.
基金supported by the Innovation Research Group Project in Universities of Chongqing of China(No.CXQT19018)the National Natural Science Foundation of China(Grant No.11971085)+1 种基金he Natural Science Foundation of Chongqing(Grant Nos.cstc2021jcyj-jqX0011 and cstc2020jcyj-msxm0777)an open project of Key Laboratory for Optimization and Control Ministry of Education,Chongqing Normal University(Grant No.CSSXKFKTM202006)。
文摘A Nitsche-based element-free Galerkin(EFG)method for solving semilinear elliptic problems is developed and analyzed in this paper.The existence and uniqueness of the weak solution for semilinear elliptic problems are proved based on a condition that the nonlinear term is an increasing Lipschitz continuous function of the unknown function.A simple iterative scheme is used to deal with the nonlinear integral term.We proved the existence,uniqueness and convergence of the weak solution sequence for continuous level of the simple iterative scheme.A commonly used assumption for approximate space,sometimes called inverse assumption,is proved.Optimal order error estimates in L 2 and H1 norms are proved for the linear and semilinear elliptic problems.In the actual numerical calculation,the characteristic distance h does not appear explicitly in the parameterβintroduced by the Nitsche method.The theoretical results are confirmed numerically。
