Iterative Solution of Large Linear Systems

PrefaceAcknowledgmentsNotationList of Fundamental Matrix PropertiesList of Iterative Methods1. Introduction 1.1. The Model Problem Supplementary Discussion Exercises2. Matrix Preliminaries 2.1. Review of Matrix Theory 2.2. Hermitian Matrices and Positive Definite Matrices 2.3. Vector Norms and Matrix Norms 2.4. Convergence of Sequences of Vectors and Matrices 2.5. Irreducibility and Weak Diagonal Dominance 2.6. Property A 2.7. L-Matrices and Related Matrices 2.8. Illustrations Supplementary Discussion Exercises3. Linear Stationary Iterative Methods 3.1. Introduction 3.2. Consistency, Reciprocal Consistency, and Complete Consistency 3.3. Basic Linear Stationary Iterative Methods 3.4. Generation of Completely Consistent Methods 3.5. General Convergence Theorems 3.6. Alternative Convergence Conditions 3.7. Rates of Convergence 3.8. The Jordan Condition Number of a 2 x 2 Matrix Supplementary Discussion Exercises4. Convergence of the Basic Iterative Methods 4.1. General Convergence Theorems 4.2. Irreducible Matrices with Weak Diagonal Dominance 4.3. Positive Definite Matrices 4.4. The SOR Method with Varying Relaxation Factors 4.5. L-Matrices and Related Matrices 4.6. Rates of Convergence of the J and GS Methods for the Model Problem Supplementary Discussion Exercises5. Eigenvalues of the SOR Method for Consistently Ordered Matrices 5.1. Introduction 5.2. Block Tri-Diagonal Matrices 5.3. Consistently Ordered Matrices and Ordering Vectors 5.4. Property A 5.5. Nonmigratory Permutations 5.6. Consistently Ordered Matrices Arising from Difference Equations 5.7. A Computer Program for Testing for Property A and Consistent Ordering 5.8. Other Developments of the SOR Theory Supplementary Discussion Exercises6. Determination of the Optimum Relaxation Factor 6.1. Virtual Spectral Radius 6.2. Analysis of the Case Where All Eigenvalues of B Are Real 6.3. Rates of Convergence: Comparison with the Gauss-Seidel Method 6.4. Analysis of the Case Where Some Eigenvalues of B Are Complex 6.5. Practical Determination of ¿b: General Considerations 6.6. Iterative Methods of Choosing ¿b 6.7. An Upper Bound for µ 6.8. A Priori Determination of µ: Exact Methods 6.9. A Priori Determination of µ: Approximate Values 6.10. Numerical Results Supplementary Discussion Exercises7. Norms of the SOR Method 7.1. The Jordan Canonical Form of L ¿ 7.2. Basic Eigenvalue Relation 7.3. Determination of ¿L ¿¿D1/2 7.4. Determination of ¿Lm ¿b¿D1/2 7.5. Determination of ¿L ¿¿A1/2 7.6. Determination of ¿Lm ¿b¿A1/2 7.7. Comparison of ¿Lm ¿b¿D1/2 and ¿Lm ¿b¿A1/2 Supplementary Discussion Exercises8. The Modified SOR Method: Fixed Parameters 8.1. Introduction 8.2. Eigenvalues of L¿, ¿1 8.3. Convergence and Spectral Radius 8.4. Determination of ¿L¿, ¿1¿D1/2 8.5. Determination of ¿L¿, ¿1¿A1/2 Supplementary Discussion Exercises9. Nonstationary Linear Iterative Methods 9.1. Consistency, Convergence, and Rates of Convergence 9.2. Periodic Nonstationary Methods 9.3. Chebyshev Polynomials Supplementary Discussion Exercises10. The Modified SOR Method: Variable Parameters 10.1. Convergence of the MSOR Method 10.2. Optimum Choice of Relaxation Factors 10.3. Alternative Optimum Parameter Sets 10.4. Norms of the MSOR Method: Sheldon's Method 10.5. The Modified Sheldon Method 10.6. Cyclic Chebyshev Semi-Iterative Method 10.7. Comparison of Norms Supplementary Discussion Exercises11. Semi-Iterative Methods 11.1. General Considerations 11.2. The Case Where G Has Real Eigenvalues 11.3.