arrow
返回

PHYSICS-BASED PRECONDITIONERS FOR OCEAN SIMULATION

delete2013-01-01
delete11
PRE
AI
C
Christopher K. Newman *
D
D. A. Knoll
DOI:10.1137/120881397delete
delete原文链接
delete原文求助
delete分享
delete收藏
摘要

摘要

En 中文
We develop and characterize a physics-based preconditioner for ocean simulation based on barotropic-baroclinic splitting. Physics-based preconditioning is a highly successful approach for multiple time scale problems where an accurate simulation is desired on the dynamical time scale. Our approach is the development of a preconditioning strategy for a fully implicit, fully coupled, second order accurate time integration of the momentum, continuity, temperature, and salinity transport equations and equation of state of ocean dynamics. The nonlinear system is solved via preconditioned Jacobian-free Newton-Krylov, where we reformulate traditional barotropic-baroclinic splitting as a preconditioner. This strategy effectively preconditions the fully coupled three-dimensional vector system by only inverting a scalar, horizontal, two-dimensional system. Furthermore, the desired solution is timestep converged with timesteps on the order of the dynamical time scale. We provide numerical examples to support the study and compare to explicit methods to verify second order convergence and demonstrate algorithmic scalability.
Keyword:
preconditioning
Jacobian-free Newton-Krylov methods
ocean modeling
AI总结

AI总结

对已上传原文的论文进行重点信息的提取,主要内容包括:简要概述、研究摘要、背景介绍、关键亮点、图文解析、展望与总结。

期刊

SIAM Journal on Scientific Computing 封面图
SIAM Journal on Scientific Computing
IF:
2.6
论文数:
5.1K
被引数:
1.8W

机构

U
united states department of energy (doe)
学者数:
11.3W
论文数: 9.6W
被引数: 246