This proposal is submitted to the FY-92 Instrumentation Competition conducted by the Division of Social and Economic Science of the National Science Foundation. It requests support to purchase a computer array for large-scale data processing and numerical analysis required by fourteen research projects. The Principal Investigators in those projects constitute a multi- disciplinary User Group from the Departments of Economics, Sociology and Psychology, and the School of Public Policy at the University of Chicago, plus the Population Research Center and Chapin Hall Center for Children at NORC. While the requesting projects represent a range of scientific interests and substantive problems, they have a unified need for similar algorithms to identify models of individual, group, or organ subsystem behavior. The commonalities can be summarized under four general headings: 1. Linear estimation using large samples; 2. Non-linear function optimization, often using large files; 3. Simulation and Monte Carlo estimation; 4. Solution of differential equation systems. Importantly, the computer array itself is an assembly of state- of-the-art components using low-cost technology with natural extensions to processing power rivaled only by super-computers. Its advantages for social science computing in cost-performance, ease-of-use, flexibility, and robustness are unmatched.
