Title :
Polynomial acceleration of Monte-Carlo global search
Author :
Calvin, James M.
Author_Institution :
Dept. of Comput. & Inf. Sci., New Jersey Inst. of Technol., Newark, NJ, USA
fDate :
6/21/1905 12:00:00 AM
Abstract :
We describe a class of algorithms for approximating the global minimum of a function defined on a subset of d-dimensional Euclidean space. The algorithms are based on adaptively composing a number of simple Monte Carlo searches, and use a memory of a fixed finite number of observations. Within the class of algorithms it is possible to obtain arbitrary polynomial speedup in the asymptotic convergence rate compared with simple Monte Carlo
Keywords :
Monte Carlo methods; convergence; function approximation; search problems; simulation; Monte-Carlo global search; asymptotic convergence rate; d-dimensional Euclidean space; function approximation; global minimum; observations; polynomial acceleration; simulation; Acceleration; Adaptive algorithm; Algorithm design and analysis; Convergence; Information science; Minimization methods; Monte Carlo methods; Polynomials; Sampling methods; Space technology;
Conference_Titel :
Simulation Conference Proceedings, 1999 Winter
Conference_Location :
Phoenix, AZ
Print_ISBN :
0-7803-5780-9
DOI :
10.1109/WSC.1999.823184
