IMDEA Software Institute researchers have presented several papers at the 34th International Conference on Logic Programming (ICLP'18), that this year was a part of the Federated Logic Conference (FLoC'18), held July 14-17, 2018, in Oxford, UK. FLoC is a major event celebrated once every four years which gathers all the major conferences related to logic and computer science. ICLP is the top international venue in the area of Logic Programming and is part of FLoC.
IMDEA Software Institute director Manuel Carro was a member of the Program Committee of ICLP this year. PhD student Joaquín Arias presented s(CASP), a system developed in collaboration with Manuel Carro and researchers from the University of Texas at Dallas. s(CASP) evaluates non-monotonic programs with constraints avoiding the combinatorial explosion due to the grounding phase. The paper entitled “Constraint Answer Set Programming without Grounding”, to be published in Theory and Practice of Logic Programming, describes this novel execution model and shows through several examples the enhanced expressiveness of s(CASP) w.r.t. ASP, CLP, and other CASP systems.
PhD student Maximiliano Klemen presented a technical communication on his work with recent PhD graduate Nataliia Stulova, and IMDEA Software faculty members Jose F. Morales, Pedro Lopez-Garcia, and Manuel Hermenegildo, entitled “Towards Static Performance Guarantees for Programs with Run-time Checks”. This work presents a framework for reasoning statically about the overhead added to programs by run-time checks in terms of both algorithmic complexity and actual cost changes. Moreover, the framework allows programmers to specify an “admissible” overhead level per program routine (function, predicate, etc.), and automatically and statically check whether the program with run-time checks adheres to such specification for all possible executions.
PhD student Isabel Garcia presented a technical communication on her work with IMDEA Software faculty members Jose F. Morales and Manuel Hermenegildo, entitled “Towards Incremental and Modular Context-sensitive Analysis”. This work tackles the issue of the high cost of context-sensitive static analysis on large code bases during the development process. The authors present an approach that combines fine-grain incrementality and modularity, which was not covered by previous approaches. Analysis results are updated at the particular places where the changes to the code are introduced (incrementality), which is significantly less costly than performing global system-level analysis. Local updates are propagated as necessary to the affected program units (modularity), obtaining the same results as standard modular global analysis with greatly reduced cost.
Faculty member Pedro Lopez-Garcia presented his work with PhD student Maximiliano Klemen, recent PhD graduate Umer Liqat, and faculty member Manuel Hermenegildo, entitled “A General Equational Framework for Static Profiling of Parametric Resource Usage” at the 19th Workshop on Logic and Computational Complexity (LCC'18). This novel framework for setting up cost relations can be instantiated for performing a wide range of resource usage analyses, including both static profiling of accumulated cost and the inference of standard notions of cost. This allows identifying statically the parts of programs that have the greatest impact on the total cost.
