27.06.2013 Views

Evolution and Optimum Seeking

Evolution and Optimum Seeking

Evolution and Optimum Seeking

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Chapter 3<br />

Hill climbing Strategies<br />

In this chapter some of the direct, mathematical parameter optimization methods will<br />

be treated in more detail for static, non-discrete, non-stochastic, mostly unconstrained<br />

functions. They come under the general heading of hill climbing strategies because their<br />

manner of searching for a maximum corresponds closely to the intuitive way a sightless<br />

climber might feel his way from a valley up to the highest peak of a mountain. For<br />

minimum problems the sense of the displacements is simply reversed, otherwise uphill or<br />

ascent <strong>and</strong> downhill or descent methods (Bach, 1969) are identical. Whereas methods of<br />

mathematical programming are dominant in operations research <strong>and</strong> the special methods<br />

of functional optimization in control theory, the hill climbing strategies are most frequently<br />

applied in engineering design. Analytic methods often prove unsuitable in this eld<br />

Because the assumptions are not satis ed under which necessary conditions for<br />

extrema can be stated (e.g., continuity of the objective function <strong>and</strong> its derivatives)<br />

Because there are di culties in carrying out the necessary di erentiations<br />

Because a solution of the equations describing the conditions does not always lead<br />

to the desired optimum (it can be a local minimum, maximum, or saddle point)<br />

Because the equations describing the conditions, in general a system of simultaneous<br />

non-linear equations, are not immediately soluble<br />

To what extent hill climbing strategies take care of these particular characteristics<br />

depends on the individual method. Very thorough presentations covering some topics can<br />

be found in Wilde (1964), Rosenbrock <strong>and</strong> Storey (1966), Wilde <strong>and</strong> Beightler (1967),<br />

Kowalik <strong>and</strong> Osborne (1968), Box, Davies, <strong>and</strong> Swann (1969), Pierre (1969), Pun (1969),<br />

Converse (1970), Cooper <strong>and</strong> Steinberg (1970), Ho mann <strong>and</strong> Hofmann (1970), Beveridge<br />

<strong>and</strong> Schechter (1970), Aoki (1971), Zahradnik (1971), Fox (1971), Cea (1971), Daniel<br />

(1971), Himmelblau (1972b), Dixon (1972a), Jacoby, Kowalik <strong>and</strong> Pizzo (1972), Stark <strong>and</strong><br />

Nicholls (1972), Brent (1973), Gottfried <strong>and</strong> Weisman (1973), V<strong>and</strong>erplaats (1984), <strong>and</strong><br />

Papageorgiou (1991). More variations or theoretical <strong>and</strong> numerical studies of older methods<br />

can be found as individual publications in a wide variety of journals, or in the volumes<br />

of collected articles such asGraves <strong>and</strong> Wolfe (1963), Blakemore <strong>and</strong> Davis (1964), Lavi<br />

23

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!