SCorES, A Logical Programming Environment for Distributed Systems

Distributed systems, important to civilian and military infrastucture, have steadily become more complex and steadily more difficult to understand, implement, and maintain. Addressing these dangers, a collaboration between ATC-NY and Cornell University will build a mathematically based tool, SCorES, providing powerful automated support for specifying, developing, verifying, and synthesizing real-time distributed systems at a high level of abstraction. Mathematical techniques for modeling and analyzing distributed systems are difficult to use because they are insufficiently abstract. SCorES supports abstract methods that are "declarative" (rather than operational) and "constructive". Declarative methods permit systems to be specified, analyzed, developed, and verified at a conceptual level congenial to human designers. Constructive methods permit automatic code synthesis. The key is to define a "logic" for this new domain, so all development steps become logical inferences. Work by Cornell and ATC-NY has already defined a logic appropriate for the class of distributed systems that can be specified and modeled without reference to quantitative real time. This logic has specified and derived demonstrably correct nontrivial distributed algorithms (e.g., consensus protocols). We will extend our methods to hybrid systems, including variables that evolve in continuous time and implement SCorES by encoding these methods within the NuPRL logical environment.