Paper 4, Section II, E

Numerical Analysis | Part II, 2015

(a) Define the mm th Krylov space Km(A,v)K_{m}(A, v) for ARn×nA \in \mathbb{R}^{n \times n} and 0vRn0 \neq v \in \mathbb{R}^{n}. Letting δm\delta_{m} be the dimension of Km(A,v)K_{m}(A, v), prove the following results.

(i) There exists a positive integer sns \leqslant n such that δm=m\delta_{m}=m for msm \leqslant s and δm=s\delta_{m}=s for m>sm>s.

(ii) If v=i=1sciwiv=\sum_{i=1}^{s^{\prime}} c_{i} w_{i}, where wiw_{i} are eigenvectors of AA for distinct eigenvalues and all cic_{i} are nonzero, then s=ss=s^{\prime}.

(b) Define the term residual in the conjugate gradient (CG) method for solving a system Ax=bA x=b with symmetric positive definite AA. Explain (without proof) the connection to Krylov spaces and prove that for any right-hand side bb the CG method finds an exact solution after at most tt steps, where tt is the number of distinct eigenvalues of AA. [You may use without proof known properties of the iterates of the CG method.]

Define what is meant by preconditioning, and explain two ways in which preconditioning can speed up convergence. Can we choose the preconditioner so that the CG method requires only one step? If yes, is it a reasonable method for speeding up the computation?

Typos? Please submit corrections to this page on GitHub.