ITR: Advanced Application Service Provider Technologies for Large-Scale Optimization

应用服务提供商（ASP）-通过互联网分发软件的使用，而不是软件本身-越来越被视为网络服务发展的下一个主要因素之一。该项目将研究和测试ASP概念在一个可能产生特别大影响的领域的应用：大规模优化软件。优化软件的复杂性和多样性对网络服务的发展提出了挑战性的障碍，但这种软件的特殊多样性和模块化确保了成功的努力将在科学、教育和商业方面产生广泛的影响。从技术上讲，该项目将建立在西北大学优化技术中心和阿贡国家实验室以前的工作基础上。一个子项目将调查资源分配的优化要求，其需求无法轻易预测。其他将涉及客户端和服务器端对象类库的设计和测试，这些对象类库在当前的数学计算中没有模型。该项目还将包括算法选择和求解器调度的自动化，通过构建和使用数据库来学习先前优化请求的结果和人类专家。因此，该项目汇集了用于计算大规模优化的方法，软件和系统，旨在使该领域令人印象深刻的技术更易于应用，算法开发和教育。

Application Service Providers (ASPs) - which distribute the use of software, rather than the software itself, via the Internet - are increasingly coming to be seen as one of the next major elements in the development of network services. This project will study and test the application of the ASP concept in an area where it is likely to have a particularly great influence: software for large-scale optimization. The complexity and variety of optimization software present challenging obstacles to the development of network services, but the exceptional variety and modularity of this software insure that successful efforts will have a broad impact in science, education, and business.Technically, the project will build on previous work by the Optimization Technology Center of Northwestern University and Argonne National Laboratory. One subproject will investigate resource allocation for optimization requests whose needs cannot be easily forecast. Others will involve the design and testing of both client-side and server-side object class libraries that have no models in current mathematical computation. The project will also include the automation of algorithm choice and solver scheduling, through the construction and use of a database to learn from results of previous optimization requests and from human experts. The project thus brings together methods, software, and systems for computational large-scale optimization, with the aim of making the impressive technology in this area more accessible for applications, algorithm development, and education.