Preconditioned CG-type methods for solving the coupled system of fundamental semiconductor equations (Q581997): Difference between revisions

From MaRDI portal
Importer (talk | contribs)
Created a new Item
 
Importer (talk | contribs)
Changed an Item
Property / review text
 
This paper presents some experimental results of applying various preconditioned conjugate gradient (CG) type methods to nonsymmetric linear equation systems arising from stationary semiconductor equations. The resultant elliptic BVP is discretized with a rectangular grid with varialble step size determined by the local gradient. The discretized equations are nonlinear, of the form \(F(v)=0\), where v is a block vector. A combimation of a continuation method and a modified Newton method is used to solve these equations, with the applied voltage as continuation parameter. The linearized systems arising from the Newton method are solved by preconditioned generalized CG methods. The authors discuss the construction of a suitable preconditioner and describe two generalized CG methods - GCGLS and CGS [cf. \textit{P. Sonneveld}, SIAM J. Sci. Stat. Comput. 10, 36-52 (1989; Zbl 0666.65029)]. Numerical results using variants of the two methods are presented and compared, together with comparisons with direct Gaussian elimination. The results indicate that a version of GCGLS is likely to be both efficient and robust for solution of the coupled system of 3 nonlinear elliptic boundary value problems describing stationary semiconductor devices.
Property / review text: This paper presents some experimental results of applying various preconditioned conjugate gradient (CG) type methods to nonsymmetric linear equation systems arising from stationary semiconductor equations. The resultant elliptic BVP is discretized with a rectangular grid with varialble step size determined by the local gradient. The discretized equations are nonlinear, of the form \(F(v)=0\), where v is a block vector. A combimation of a continuation method and a modified Newton method is used to solve these equations, with the applied voltage as continuation parameter. The linearized systems arising from the Newton method are solved by preconditioned generalized CG methods. The authors discuss the construction of a suitable preconditioner and describe two generalized CG methods - GCGLS and CGS [cf. \textit{P. Sonneveld}, SIAM J. Sci. Stat. Comput. 10, 36-52 (1989; Zbl 0666.65029)]. Numerical results using variants of the two methods are presented and compared, together with comparisons with direct Gaussian elimination. The results indicate that a version of GCGLS is likely to be both efficient and robust for solution of the coupled system of 3 nonlinear elliptic boundary value problems describing stationary semiconductor devices. / rank
 
Normal rank
Property / Mathematics Subject Classification ID
 
Property / Mathematics Subject Classification ID: 65F10 / rank
 
Normal rank
Property / Mathematics Subject Classification ID
 
Property / Mathematics Subject Classification ID: 65N30 / rank
 
Normal rank
Property / Mathematics Subject Classification ID
 
Property / Mathematics Subject Classification ID: 65Z05 / rank
 
Normal rank
Property / Mathematics Subject Classification ID
 
Property / Mathematics Subject Classification ID: 35Q99 / rank
 
Normal rank
Property / Mathematics Subject Classification ID
 
Property / Mathematics Subject Classification ID: 78A55 / rank
 
Normal rank
Property / zbMATH DE Number
 
Property / zbMATH DE Number: 4129877 / rank
 
Normal rank
Property / zbMATH Keywords
 
preconditioned conjugate gradient method
Property / zbMATH Keywords: preconditioned conjugate gradient method / rank
 
Normal rank
Property / zbMATH Keywords
 
experimental results
Property / zbMATH Keywords: experimental results / rank
 
Normal rank
Property / zbMATH Keywords
 
nonsymmetric linear equation systems
Property / zbMATH Keywords: nonsymmetric linear equation systems / rank
 
Normal rank
Property / zbMATH Keywords
 
stationary semiconductor equations
Property / zbMATH Keywords: stationary semiconductor equations / rank
 
Normal rank
Property / zbMATH Keywords
 
continuation method
Property / zbMATH Keywords: continuation method / rank
 
Normal rank
Property / zbMATH Keywords
 
Newton method
Property / zbMATH Keywords: Newton method / rank
 
Normal rank
Property / zbMATH Keywords
 
Numerical results
Property / zbMATH Keywords: Numerical results / rank
 
Normal rank
Property / zbMATH Keywords
 
comparisons
Property / zbMATH Keywords: comparisons / rank
 
Normal rank
Property / zbMATH Keywords
 
Gaussian elimination
Property / zbMATH Keywords: Gaussian elimination / rank
 
Normal rank

Revision as of 18:01, 1 July 2023

scientific article
Language Label Description Also known as
English
Preconditioned CG-type methods for solving the coupled system of fundamental semiconductor equations
scientific article

    Statements

    Preconditioned CG-type methods for solving the coupled system of fundamental semiconductor equations (English)
    0 references
    0 references
    0 references
    0 references
    1989
    0 references
    This paper presents some experimental results of applying various preconditioned conjugate gradient (CG) type methods to nonsymmetric linear equation systems arising from stationary semiconductor equations. The resultant elliptic BVP is discretized with a rectangular grid with varialble step size determined by the local gradient. The discretized equations are nonlinear, of the form \(F(v)=0\), where v is a block vector. A combimation of a continuation method and a modified Newton method is used to solve these equations, with the applied voltage as continuation parameter. The linearized systems arising from the Newton method are solved by preconditioned generalized CG methods. The authors discuss the construction of a suitable preconditioner and describe two generalized CG methods - GCGLS and CGS [cf. \textit{P. Sonneveld}, SIAM J. Sci. Stat. Comput. 10, 36-52 (1989; Zbl 0666.65029)]. Numerical results using variants of the two methods are presented and compared, together with comparisons with direct Gaussian elimination. The results indicate that a version of GCGLS is likely to be both efficient and robust for solution of the coupled system of 3 nonlinear elliptic boundary value problems describing stationary semiconductor devices.
    0 references
    preconditioned conjugate gradient method
    0 references
    experimental results
    0 references
    nonsymmetric linear equation systems
    0 references
    stationary semiconductor equations
    0 references
    continuation method
    0 references
    Newton method
    0 references
    Numerical results
    0 references
    comparisons
    0 references
    Gaussian elimination
    0 references

    Identifiers

    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references