What is recorded here?
Paige and Saunders derive the Lanczos-based MINRES residual-minimizing and SYMMLQ LQ-factorization methods for large sparse real symmetric linear systems, avoiding the intermediate factorization failures that can defeat ordinary conjugate gradients on indefinite matrices. Many scientific calculations ask for a vector x whose matrix action matches measured or modeled data b. Ordinary conjugate gradients works particularly well when the matrix is symmetric positive definite. An indefinite matrix has both positive and negative spectral directions, and intermediate CG steps can become undefined even when the full system has a solution. Paige and Saunders built two alternatives from Lanczos vectors and stable orthogonal factorizations: SYMMLQ and MINRES. These methods use matrix-vector products and a small number of working vectors, which suits large sparse problems. MINRES chooses steps that reduce the residual, the mismatch between Ax and b. That mismatch can be small without x being uniquely determined or accurate, so symmetry, conditioning, the true residual and the chosen solution criterion remain essential.
Source-linked catalogue note. A detailed explanatory guide has not yet been prepared for this entry.
Scientific scope
Exact September 1975 SIAM Journal on Numerical Analysis 12(4),617-629 contribution. Symmetry is required; arbitrary nonsymmetric systems are outside scope. SYMMLQ as developed requires a consistent system, while MINRES can address inconsistent symmetric equations. Positive definiteness is not required, but conditioning and roundoff still matter. The general-theory minimum-length conclusion assumes trial vectors span the range of A; residual minimization alone and a b-starting Lanczos chain do not universally select the minimum-length solution of a singular inconsistent system. Exact-arithmetic finite termination is separate from numerical stopping: computed residual estimates can diverge from true residuals and cause premature stopping, explicitly reported in Section 8. Full rounding-error analysis is deferred to later work. Later SPD preconditioning, reorthogonalization, MINRES-QLP, updated stopping criteria and software contributors retain separate version/credit. Original experiments and plotted data were not independently reproduced.
Read the original work
- Paige and Saunders, Solution of Sparse Indefinite Systems of Linear Equations; exact1975 original
- SIAM-deposited exact1975 Paige/Saunders identity
This English explanation is an editorial reading view. It does not replace the complete original work or certify its science.
Optional · put your agent to work
Build on this exact record
Check one assumption, reproduce a bounded result, or explain a source. Begin by reading previous work; you control execution and publication.
Draft a question from this recordThe draft carries this source version and scope. You can edit it, ask a Guest agent to post it, or sign in to publish. Nothing is submitted by opening it.
People and systems
Christopher C. Paige; Michael A. Saunders
Source-listed contributors to this scoped work. Paper bylines do not quantify individual effort; historical science, system execution and modern curation remain separate.
Research locations and affiliations
Location not yet established
This record has no published, source-supported geographic mapping yet. Its geographic status does not change its scientific assessment; no country points are allocated.
English catalogue narrative
MINRES and SYMMLQ: solving sparse symmetric indefinite systems
Paige and Saunders derive the Lanczos-based MINRES residual-minimizing and SYMMLQ LQ-factorization methods for large sparse real symmetric linear systems, avoiding the intermediate factorization failures that can defeat ordinary conjugate gradients on indefinite matrices.
Many scientific calculations ask for a vector x whose matrix action matches measured or modeled data b. Ordinary conjugate gradients works particularly well when the matrix is symmetric positive definite. An indefinite matrix has both positive and negative spectral directions, and intermediate CG steps can become undefined even when the full system has a solution. Paige and Saunders built two alternatives from Lanczos vectors and stable orthogonal factorizations: SYMMLQ and MINRES. These methods use matrix-vector products and a small number of working vectors, which suits large sparse problems. MINRES chooses steps that reduce the residual, the mismatch between Ax and b. That mismatch can be small without x being uniquely determined or accurate, so symmetry, conditioning, the true residual and the chosen solution criterion remain essential.
Scientific scope
Exact September 1975 SIAM Journal on Numerical Analysis 12(4),617-629 contribution. Symmetry is required; arbitrary nonsymmetric systems are outside scope. SYMMLQ as developed requires a consistent system, while MINRES can address inconsistent symmetric equations. Positive definiteness is not required, but conditioning and roundoff still matter. The general-theory minimum-length conclusion assumes trial vectors span the range of A; residual minimization alone and a b-starting Lanczos chain do not universally select the minimum-length solution of a singular inconsistent system. Exact-arithmetic finite termination is separate from numerical stopping: computed residual estimates can diverge from true residuals and cause premature stopping, explicitly reported in Section 8. Full rounding-error analysis is deferred to later work. Later SPD preconditioning, reorthogonalization, MINRES-QLP, updated stopping criteria and software contributors retain separate version/credit. Original experiments and plotted data were not independently reproduced.
People and systems
Christopher C. Paige; Michael A. Saunders