On descent from local minima
Abstract: When a local minimum of a function of several variables has been found by use of an algorithm for finding such minima numerically, one often runs the same algorithm many times with different starting values in the hopes of finding a lower minimum. Here, under the assumption that a local minimum is known, a process with analytical criteria is described which sometimes finds smaller local minima in an algorithmic manner.
Retrieve articles in Mathematics of Computation with MSC: 49D05
Retrieve articles in all journals with MSC: 49D05
Keywords: Descent methods, global minimization, local minima, numerical methods, unconstrained minimization
Article copyright: © Copyright 1971 American Mathematical Society