Topics in the Analysis and Optimization of Complex Systems.
Abstract
This document is a report of the results of research under AFOSR Contract No. F44620-71-C-0093 through Dec. 1975. The general thrust of the research is the development of approaches to improving the utility of computer-oriented quantitative methods in practical decision-making and system design. The research was directed at obtaining acceptable system models in two ways: (1) by repro modeling (making a more efficient model of) a complex model based on physical processes; and (2) by developing empirical models based on limited measurements of the system. Both problems are abstractly identical: In the first, case, the data is obtained from a model; in the second case, from measurements on the system in question or similar systems. The abstract problem in most applications of interest is the analysis and efficient representation of the relationship between multiple input variables or parameters and an output variable. The number of input/output samples frequently large in an absolute sense, but generally samll relative to their dimensionality. Usually the data if essentially nonlinear with complex interactions between variables. Classical statistical methods of analysis are largely inapplicable. One main effort over the contract period was in developing and extending methods designed for this type of data structures. Another effort in a parallel direction was the development of effective decision procedures based on model outputs. This resulted in methods for dealing with decision making in multicriterion situations, and in the study of tree structured decision procedures. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1976
- Accession Number
- ADA023181
Entities
People
- Leo Breiman
- William S. Meisel
Organizations
- Technology Service Corporation