nlsimplex
non-linear simplex minimization
Description
Find the minimum point of a function by the non-linear simplex method.
Usage
nlsimplex(func, pars, max.iter=500)
Required Arguments
- func
-
The name of the function to be minimized.
The function func must take a single, vector argument of the parameters for the minimization.
- pars
-
The argument to func indicating the starting point for iterations.
Alternatively, para can be a list as returned by nlsimplex, which can be used for continuation.
Optional Arguments
- max.iter
-
The maximum number of iterations.
Value
A list of: $simplex.values - the function values for points in the simplex arranged in
increasing order, $simplex.parms the array of simplex points, $converged - a logical indication
of convergence (T means convergered), $value indicating the minimum, and
$parms indicating the parmameters at the minimum.
.
Details
Procedure nlsimplex uses the Nelder-Mead algorithm to find a set of
parameters that give a (local) minimum. For maximization (eg. the
likelihood function) simply try to find the minimum of the negation of
the function. The algorithm does not require that the function is
differentiable, or even continuous, so it is quite robust, but can be
very slow. Each iteration is very simple, but the number of iteration
may need to be very large. This version of nlsimplex reflects away
from the worst point.
This program does not seem to converge on moderately difficult problems.
REFERENCES
For more information about the algorithm see Numerical Recipes,
William H. Press, B.P.Flannery, S.A. Teukolsky and W.T.Vetterling, p 289.
See Also
Examples
return to Table of Contents