AKL(ID:2921/akl002)

Agents Kernel Language 

for Agents Kernel Language

Intelligent Systems Laboratory
Swedish Institute of Computer Science.
Concurrent, constraint programming language. Computation is performed by agents interacting through stores of constraints. This notation accommodates multiple programming paradigms; in appropriate contexts, AKL agents may be thought of as processes, objects, functions, relations, and constraints.

Structures:
Related languages
AKL => AKL   Renaming
AKL => Penny   Extension of
References:
  • Sahlin, Dan and Thomas Sjöland. Abstract Syntax and Normalisation of AKL Programs. Appendix to deliverable D.WP1.6.1.M1 in the ESPRIT project ParForce, 6707, 1993. view details
  • Sahlin, Dan and Thomas Sjöland. Towards an Analysis Tool for AKL. Deliverable D.WP1.6.1.M1 in the ESPRIT project ParForce, 6707, 1993. view details
  • Brand, Per Enhancing the AKL compiler using global analysis. Deliverable D.WP2.1.3.M2 in the ESPRIT project ParForce, 6707, 1994. view details
  • Carlson, B.; Carlsson, M.; and D. Diaz. Entailment of Finite Domain Constraints" view details
          in Proceedings of the Eleventh International Conference of Logic Programming, MIT Press Series in Logic Programming, 1994 view details
  • Carlson, B.; Janson, S. ; and Haridi, S. "AKL(FD): A concurrent language for FD programming" view details
          in Proceedings of the 1994 International Logic Programming Symposium, MIT Press Series in Logic Programming, 1994. view details
  • Janson, Sverker AKL - A Multiparadigm Programming Language. SICS Dissertation Series 14. Uppsala Theses in Computing Science 19. Uppsala University, Sweden, June 1994. ISBN 91-506-1046-5. view details
          in Proceedings of the 1994 International Logic Programming Symposium, MIT Press Series in Logic Programming, 1994. view details
  • PARallel FORmal Computing Environment (ESPRIT Basic Research) EATCS 1994 Report view details Abstract: ParForCE is aimed at constructing (and evaluating the use of) formal tools for the development of parallel programs and their efficient execution. To this end the emerging techniques for formal program analysis and manipulation are applied to central issues relating to parallel execution such as dependency and granularity analysis or memory management. Tools based on these techniques are built to aid in the formal development of parallel logic programs. These tools are then integrated with parallel execution platforms and their effectiveness assessed. Extract: Parallel Logic Programming Platforms
    A number of parallel logic programming systems, which have been shown to be capable of significant speedups over state of the art sequential Prolog systems, are already being used as platforms by the project partners. These include &-Prolog [39] (developed at Madrid), Muse [2, 3] (developed at SICS), and ElipSys [7] (developed at ECRC). The ElipSys execution platform, a prototype of which was developed as part of ESPRIT Project 2025, EDS, supports OR-parallelism as well as a framework for implementing constraint solvers and a close coupling to databases. Muse is a well established execution platform that allows Prolog to be executed in parallel by exploiting OR-parallelism. The &-Prolog system is also a well established parallel Prolog system that exploits (independent) AND-parallelism. Some parts of the &-Prolog and Muse models were developed by SICS and Madrid in ESPRIT Project 2471, PEPMA.

    A matter of great interest is the combination of the capabilities of these systems, as well as those of other languages, both from the point of view of concurrency and constraint support.

    The AGENTS Language (AKL) is a new concurrent logic language developed by SICS in ESPRIT Project 2471, PEPMA, with a large potential for parallel execution. It provides the programming paradigms of search-oriented languages such as Prolog, process-oriented languages such as GHC, and the constraint logic programming languages in a unified framework.

    A proposal for combining Muse and &-Prolog into one system to exploit their two sources of parallelism while maintaining the high efficiency of both systems is represented by the ACE model [35]. We refer to a further enhancement of this model which also supports constraint solving (and explicit concurrency) in the same framework as Ciao-Prolog --(Concurrent,) Constraint, Independence-based And/Or parallel Prolog.

          in Proceedings of the 1994 International Logic Programming Symposium, MIT Press Series in Logic Programming, 1994. view details
  • Sahlin, Dan and Thomas Sjöland. Fixpoint Analysis of Type and Alias in AKL programs. Deliverable D.WP1.6.1.M2 in the ESPRIT project ParForce, 6707, 1994. view details
          in Proceedings of the 1994 International Logic Programming Symposium, MIT Press Series in Logic Programming, 1994. view details
  • Sahlin,.Dan Specification of a Partial Evaluator for AKL. Deliverable D.WP1.4.2.M2 in the ESPRIT project ParForce, 6707, 1994. view details
          in Proceedings of the 1994 International Logic Programming Symposium, MIT Press Series in Logic Programming, 1994. view details
  • Sjöland, Thomas and Dan Sahlin. Fixpoint Analysis of Types and Alias in AKL Programs. SICS report R94:13b. 1994 view details Abstract: We have defined and implemented a method for analysis of the CCP language AKL in the spirit of abstract interpretation that uses a static set of semantic equations which abstracts the concurrent execution of an AKL program. The method strictly separates the setting up of the equation system from the solving of the system with a fixpoint procedure. The computation strategies used, results for a number of test programs and the conclusions we draw from this experimental effort are reported. The software implementing the system described herein, is deliverable number D.WP.1.6.1.M2 in the ESPRIT project ParForce 6707.
          in Proceedings of the 1994 International Logic Programming Symposium, MIT Press Series in Logic Programming, 1994. view details
  • Torkel Franzén Some Formal Aspects of AKL SICS R94-10 view details Abstract: The Agents Kernel Language allows committed choice programming as well as nondeterministic logic programming based on the Andorra principle (and combinations of these), with a considerable degree of parallelism. In the present report the AKL computation model is formally defined and soundness and completeness results for a logical subset of AKL are presented. The computation model is extended to cover ports, a medium of communication used in AKL, and the solution collecting operation bagof. These extensions preserve the basic character of the computation model, and in particular the role played by the constraint theory. Declarative interpretations of these constructs are introduced, and the limitations of these readings are discussed. Finally, the confluence of strongly fair (possibly infinite) computations in a different subset of AKL is proved.

          in Proceedings of the 1994 International Logic Programming Symposium, MIT Press Series in Logic Programming, 1994. view details
    Resources