Global computing by networked equational constraint solvers

我们致力于设计和实现一个用于协作约束求解的开放系统。我们努力的结果是一个称为Open CFLP的系统，它为开放环境中的协作约束功能逻辑编程提供了支持。该系统提供了以下支持：(a)配备了专门求解器的约束域上的高阶函数逻辑编程。功能逻辑程序的推理由基于高阶惰性缩窄演算的专用求解器实现；(b)通过专门代理的查找服务透明地访问专门的约束求解器；C一种协作语言，它使用户能够指定约束求解器协作以实现预期结果的最常用方式。对于(a)，我们在Mathematica中设计并实现了与编程目的相关的高阶惰性窄化的各种改进。这些改进非常重要，因为它们减少了解决方案的巨大搜索空间。对于(b)，我们设计并实现了一个查找服务，该服务允许透明地访问由提供商在开放环境（如Internet）中发布的约束求解器。用户请求以特定接口和属性（例如，约束求解域、求解方法等）为特征的求解器，查找服务生成这样一个远程求解器的代理（如果可用）。对于c语言，我们设计并实现了求解器协作语言，该语言提供了描述将组件求解器组合为更高级别约束求解过程的最常见方法的原语。

We worked on the design and implementation of an open system for collaborative constraint solving. The outcome of our efforts is a system called Open CFLP, which provides support for collaborative constraint functional logic programming in open environments. The system provides support for(a) higher-order functional logic programming over constraint domains equipped with specialized solvers. Reasoning over functional logic programs is realized by specialized solvers based on higher-order lazy narrowing calculi ;(b) transparent access to specialized constraint solvers via the lookup service of a specialized broker ;c a collaboration language which enables the user to specify the most common ways in which constraint solvers should collaborate to achieve the desired results.For (a), we have designed and implemented in Mathematica various refinements of higher-order lazy narrowing which are relevant for programming purposes. These refinements are very important because they reduce the huge search space for solutions. For (b), we have designed and implemented a lookup service that allows transparent access to constraint solvers advertised by providers in an open environment such as the Internet. The user asks for a solver characterized by a certain interface and attributes (e.g., constraint solving domain, solving method, etc.) and the lookup service yields a proxy to such a remote solver (if available). For c, we have designed and implemented a solver collaboration language which provides primitives to describe the most common ways to combine component solvers into a higher-level constraint solving procedure.