steep

Steepest ascent/descent is a simple and efficient optimization method ...

This m-file provides a simple and efficient optimization method based on statistical design of experiments by the steepest ascent/descent procedure to predict points headed (hopefully) toward to optimum (maximum or minimum) for a first-order design.

Inputs:
B - vector of regression coefficients (parameters) of the fitted linear equation (first order): [intercept (1), linear (p)]. It must to be enter in that strictly order.
V - vector of the factor values at starting point (current factor setting).
C - vector of units increment chosen between the low and medium levels and between the medium and high levels for each parameter.
s - number of interested steps.
o - change relative to one unit change regarding to the maximum absolute parameter [o = 1 (default)] or parameter choice by experimenter (o = 2).
d - steepest ascent [d = 1 default)] or descent (d = 2) procedure.

Outputs:
A complete summary of the estimation of the selected direction of the steepest method.