A Monte Carlo simulated annealing approach to optimization over continuous variables. (English) Zbl 0551.65045

Vanderbilt, David; Louie, Steven G.

J. Comput. Phys. 56, 259-271 (1984).

Numerical optimization methods based on thermodynamic concepts are extended to the case of continuous multidimensional parameter spaces. Techniques which allow this strategy to be implemented efficiently and reliably, including a self-regulatory mechanism for choosing the random step distribution, are described. The method is applied to a set of standard global minimization problems, and to a typical non-linear least- squares functional fitting problem.

MSC:

65K05	Numerical mathematical programming methods
65D10	Numerical smoothing, curve fitting
65C05	Monte Carlo methods

Keywords:

combinatorial optimization; thermodynamic concepts; random step distribution; global minimization; non-linear least-squares functional fitting problem

References:

[1]	Kirkpatrick, S.; Gelatt, C.D.; Vecchi, M.P., Optimization by simulated annealing, Science, 220, 671, (1983) · Zbl 1225.90162
[2]	Metropolis, N.; Rosenbluth, A.; Rosenbluth, M.; Teller, A.; Teller, E., J. chem. phys., 21, 1087, (1953)
[3]	Dahlquist, G.; Bjoerck, A., Numerical methods, (1974), Prentice-Hall Englewood Cliffs, N.J, See, for example
[4]	Dixon, L.C.W.; Szego, G.P., (), 1-18
[5]	Newer, J.A.; Mead, R., Comput. J., 7, 308, (1967)
[6]	Torn, A.A., (), 49-62
[7]	Gomulka, J., (), 63-70
[8]	Price, W.L., (), 71-84
[9]	Mockus, J.; Tiesis, V.; Zilinskas, A., (), 117-130
[10]	De Biase, L.; Frontini, F., (), 85-102
[11]	Hamann, D.R.; Schluter, M.; Chiang, C., Phys. rev. lett., 43, 1494, (1979)