Iterative Solution of Large Sparse Systems of Equations by Wolfgang Hackbusch

By Wolfgang Hackbusch

In the second one variation of this vintage monograph, entire with 4 new chapters and up to date references, readers will now have entry to content material describing and analysing classical and smooth equipment with emphasis at the algebraic constitution of linear new release, that is often neglected in different literature.
The priceless volume of labor raises dramatically with the scale of platforms, so one has to go looking for algorithms that the majority successfully and appropriately resolve platforms of, e.g., numerous million equations. the alternative of algorithms will depend on the exact homes the matrices in perform have. an immense category of huge platforms arises from the discretization of partial differential equations. for that reason, the matrices are sparse (i.e., they include in most cases zeroes) and well-suited to iterative algorithms.
The first variation of this e-book grew out of a chain of lectures given through the writer on the Christian-Albrecht collage of Kiel to scholars of arithmetic. the second one variation comprises really novel approaches.

Show description

Read or Download Iterative Solution of Large Sparse Systems of Equations (Applied Mathematical Sciences) PDF

Best mathematics_1 books

Arithmétique et travaux pratiques cycle d'observation classe de sixième

Manuel de mathématiques, niveau sixième. Cet ouvrage fait partie de l. a. assortment Lebossé-Hémery dont les manuels furent à l’enseignement des mathématiques ce que le Bled et le Bescherelle furent à celui du français.

Extra info for Iterative Solution of Large Sparse Systems of Equations (Applied Mathematical Sciences)

Example text

Square matrices are mappings of the set I × I of index pairs into K. The set of these matrices is denoted by KI×I . Matrices are always symbolised by uppercase letters. The matrix entry of A corresponding to the index pair (α, β) ∈ I × I is written as aαβ or aα,β , and occasionally as Aαβ . Alternatively, the notation A[α, β] = a[α, β] is used. In particular, (A + B)αβ , (A−1 )αβ , etc. is written for the components of matrix expressions. The matrix composed of the entries aαβ is denoted by A = (aαβ )α,β∈I .

This property is the subject of the following definition. 5 (consistency). 1) with A ∈ D(Φ) if, for all right-hand sides b ∈ KI , any solution of Ax = b is a fixed point of Φ(·, b, A). 5, consistency means: For all b, x ∈ KI and all matrices A ∈ D(Φ), the implication Ax = b ⇒ x = Φ(x, b, A) holds. The reverse implication would yield an alternative (nonequivalent) form of consistency: Ax = b for all fixed points x of Φ(·, b, A) and for all b ∈ KI, A ∈ D(Φ). 2). 2), there may be a solution of Ax = b for certain b.

The same effect is obtained by the formal definition ui+p,j+q := 0. 14) h−2 ⎣ −1 4 −1 ⎦ . 14)) are defined by zero. 4 A Single System Versus a Family of Systems Usually, a discretisation matrix A is embedded into a family {Ah }h∈H . Here H is an infinite set with accumulation point 0 ∈ H. For instance, the Poisson model problem is defined for all N ∈ N\{1} and the corresponding step sizes h := N1 → 0. , ≤ 1 − Chκ for some fixed C). Such expressions only make sense if there is a limit process h → 0. Given a family {Aη }η∈F of matrices, one is interested in the behaviour of the convergence rates (or of the computational cost for obtaining a certain accuracy) with respect to a limit process η → 0 (or η → ∞).

Download PDF sample

Rated 4.32 of 5 – based on 16 votes