Abstract

Algorithms for solving nonlinear systems of equations can be broadly classified as (1) locally convergent or (2) globally convergent. The former includes Newton's method, various quasi-Newton methods, and inexact Newton methods. The latter includes continuation, simplicial methods, and probability-one homotopy methods. These algorithms are qualitatively significantly different, and their relative performance on parallel systems may very well be the reverse of their performance on serial processors.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call