The Convergence of a Parallel Analogue of the Bisection Method.
Abstract
The number of iterations required for convergence of a parallel analogue of the bisection method for locating real zero of a function is given. The algorithm was first described by G. S. Shedler. At each iteration the algorithm evaluates the function at N equally spaced points, assigning one evaluation to each of N processors working in parallel. For the next iteration one of the N+1 subintervals is selected. Starting with an interval of length d which contains the zero, the algorithm requires (log to the base (N+1) (d/epsilon) iterations for an accuracy of epsilon in the computed zero.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 01, 1973
- Accession Number
- AD0769112
Entities
People
- Ping Y. Lam
- William G. Poole Jr.
Organizations
- College of William & Mary