Residual time extrapolation operators for efficient wavefield construction
Author(s) -
Tariq Alkhalifah
Publication year - 2012
Publication title -
king abdullah university of science and technology repository (king abdullah university of science and technology)
Language(s) - English
Resource type - Conference proceedings
DOI - 10.1190/segam2012-0217.1
Subject(s) - extrapolation , residual , discretization , computer science , computational science , homogeneous , stability (learning theory) , algorithm , mathematics , physics , statistical physics , mathematical analysis , machine learning
Solving the wave equation using nite-di erence approximations su ers from dispersion and stability related issues that might limit e cient or proper extrapolation of high frequencies. Spectral-based time extrapolation methods tend to mitigate these problems, but at an additional cost to the extrapolation. I investigate the prospective of using a residual formulation of the spectral approach, along with utilizing Shanks transform based expansions, that adheres to the residual requirements, to improve accuracy and reduce the cost. Utilizing the fact that spectral methods excel (time steps are allowed to be large) in homogeneous and smooth media, the residual implementation allows for velocity discretization that optimizes the use of this feature.
Accelerating Research
Robert Robinson Avenue,
Oxford Science Park, Oxford
OX4 4GP, United Kingdom
Address
John Eccles HouseRobert Robinson Avenue,
Oxford Science Park, Oxford
OX4 4GP, United Kingdom