
The remaining blocks in the flow chart are for computing the objective function, the gradient vector,
and for determining how big a step to take in the negati ve gradient direction in searching for a
minimum, that is, points where F(F
1
, c) ¼0.
The so-called steepest descent gradient searches (Wilde 1964) look for the optimum distance to
travel in the negative gradient direction before changing directions. The optimum distance is
determined by the local minimum of the objective function along the negative gradient direction.
When the local minimum is reached, the gradient vector is recalculated, and the search proceeds in the
new direction that happens ...