Minimization Algorithms for Functions with Random Noise

SYNOPTIC ABSTRACT

Many physical systems are functions of multiple factors too complex to model explicitly. As a result, the functional model relating factors to a particular attribute includes a random term for unmodeled effects. We examine the effectiveness of four optimization strategies for this problem class, using Monte Carlo techniques.

Key Words and Phrases:

• stochastic optimization
• performance testing
• unconstrained minimization
• Monte Carlo analysis