Tree relaxation: a new iterative solution method for linear equations

Author

Shi, Chuanjin ; Zhang, Kaihe

Author_Institution

Dept. of Electron. Eng., Fudan Univ., Shanghai, China

fYear

1988

fDate

7-9 June 1988

Firstpage

2355

Abstract

In the tree relaxation (TR) method, a network is first partitioned into a tree and links. The portions of the equations associated with links are relaxed, and the portions associated with the tree can be solved efficiently with an incremental linear algorithm. It has been shown that the proposed relaxation method has better convergence property than Gauss-Jacobi and Gauss-Seidel methods for the tree-structure dominant network. TR encounters no special problems with floating elements, which provides a way to deal with the floating-capacitance problem in conventional relaxation-based circuit simulation.<>

Keywords

circuit analysis computing; field effect integrated circuits; large scale integration; Gauss Jacobi methods; Gauss-Seidel methods; convergence; floating elements; floating-capacitance; incremental linear algorithm; iterative solution method; linear equations; relaxation method; relaxation-based circuit simulation; tree relaxation; tree-structure dominant network; Capacitance; Circuit simulation; Convergence; Equations; Gaussian processes; Iterative algorithms; Iterative methods; Partitioning algorithms; Tree data structures; Voltage;

fLanguage

English

Publisher

ieee

Conference_Titel

Circuits and Systems, 1988., IEEE International Symposium on

Conference_Location

Espoo, Finland

Type

conf

DOI

10.1109/ISCAS.1988.15416

Filename

15416