Next-Generation Servers for Optimization as an Internet Resource

点击翻译按钮获取中文摘要

The primary impediment to broader use of optimization models and methods today is one of communication. Increasing numbers of solvers are implemented increasingly well, but prospective users are unaware of them or do not see the benefit that would justify installing them. Modeling systems also tend to be slow to support new solvers, especially ones that address new problem types. Sending optimization problems over the Internet, to a solver at some remote site, is an increasingly practical way of addressing communication problems in large-scale optimization. A remote optimization server can accommodate numerous problem types and can provide varied solvers for problems of each type, giving modelers much more of a choice than they have locally. Yet current optimization servers only begin to address the communication challenges of large-scale optimization with respect to solver choice, scheduling, benchmarking, and modeling system interaction, and are limited by early design decisions. The planned research is motivated by a vision of a next-generation server that addresses outstanding challenges of communication in large-scale optimization. This work will address design as well as implementation issues posed by standardizing problem representations, automating problem analysis and solver choice, working with new web-service standards, scheduling computational resources, benchmarking solvers, and verification of results - all in the context of the special requirements of large-scale optimization. The Network Enabled Optimization System (NEOS) project seeks to make optimization a part of the worldwide software infrastructure that supports science, engineering, and commerce. Optimization analyzes complex cost-minimization and profit-maximization problems that involve large numbers of distinct but interrelated decisions. The analysis is accomplished by representing problems as mathematical models that are communicated with data to a variety of computational solvers. The NEOS Server provides remote access to about 50 solvers and hence is the focus of the planned research. The ready availability of optimization tools has widespread benefits, both directly to practitioners, and indirectly by improving the quality of research and education in optimization techniques. By encouraging comparison between solvers, moreover, the NEOS Server's variety of solvers and interfaces also serves as a spur to further improvement of optimization software.