A new multilevel Green's function interpolation method for large-scale low-frequency EM simulations

In this paper, a new multilevel Green's function interpolation method (MLGFIM) is presented to solve integral equations for large-scale electrostatic problems. In MLGFIM, the problem domain is first divided into multilevel cubes. Next, the peer-level Green's function interpolation techniqu...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on computer-aided design of integrated circuits and systems 2005-09, Vol.24 (9), p.1427-1443
Main Authors: Wang, Hao Gang, Chan, Chi Hou, Tsang, Leung
Format: Article
Language:English
Subjects:
Capacitance
Computational complexity
Computational modeling
Electrostatics
Green's function methods
Integral equations
Interpolation
Large-scale systems
microelectromechanical system
Microelectromechanical systems
multilevel Green's function interpolation method (MLGFIM)
Radiofrequency integrated circuits
RFIC
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In this paper, a new multilevel Green's function interpolation method (MLGFIM) is presented to solve integral equations for large-scale electrostatic problems. In MLGFIM, the problem domain is first divided into multilevel cubes. Next, the peer-level Green's function interpolation technique is employed, and then, a new lower-to-upper-level Green's function interpolation technique is devised. They are used with the multilevel discretization to speed up the matrix-vector multiplications in the iterative solution in which a computational complexity of O(N) is achieved. The MLGFIM is used to extract the capacitances encountered in radio frequency integrated circuits (RFICs) and microelectromechanical systems. Moreover, to demonstrate its efficiency both in simulation speed and memory storage requirement, MLGFIM is compared with FastCap for free space problems and applied to extract capacitances from multilayered structures. For problems with 375,000 unknowns, the proposed method only requires 343 MB of computer memory.
ISSN:0278-0070
1937-4151
DOI:10.1109/TCAD.2005.850804