Developing and Understanding Methods for Large-Scale Nonlinear Optimization
Abstract
The most important contribution under this grant has been a vast improvement in the ability of large-scale global optimization methods to solve very difficult molecular configuration problems. The improvements are shown both in the size and in the complexity of the problems that can now be solved. At the beginning of this research period, we and others were just beginning to solve molecular cluster problems with up to hundreds of parameters. By the end of the research period, we had progressed from these cluster problems to much more difficult and important protein folding problems with thousands of variables. The most fundamental research advance that contributed to the improvements in our large-scale global optimization algorithms was the development and understanding of new smoothing approaches. Other significant research contributions included: the successful application of our global optimization approach to distance geometry problems; the development of a new primal-dual interior point method for nonlinearly constrained optimization; and the modification and improvement of our widely used modified Cholesky factorization method.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 16, 1999
- Accession Number
- ADA369917
Entities
People
- Richard H. Byrd
- Robert B. Schnabel
Organizations
- University of Colorado Boulder