Etude et mise en oeuvre de la réécriture conditionnelle concurrente sur des machines parallèles à mémoire distribuée

Abstract : Term Rewriting is a well-known concept used in many programming languages such OBJ, as well as a fundamental mechnaism for automated deduction and formal computation. This thesis is an approach to the definition and the implementation of a conditional concurrent model on distributed memory machines. Our approach is to use the data parallelism present in term structure. Concurrent rewriting consists in the simultaneous application of several rewriting rules on the same term. First, we describe the formalisation of our model in the context of J. Meseguer's rewriting logic. To extend the model to conditional rewriting, we define a transformation from conditional to unconditional rewriting systems. We detail the implementation model and we describe the transition from the conditional concurrent rewriting abstract model to an operational model which can be implemented on distributed memory machines. The implemented model is based on asynchronous exchange of messages. We propose a concurrent garbage algorithm reclaiming unused nodes of the graph. The ReCo system implements the oerational model using MPI ou PVM message passing interface library. To validate and to experiment our implementation, different benchmarks for computing term normal forms are realised.
Document type :
Theses
File URL :
http://docnum.univ-lorraine.fr/prive/SCD_T_1997_0011_ALOUINI.pdf
Complete list of metadatas

https://hal.univ-lorraine.fr/tel-01747447
Contributor : Thèses Ul <>
Submitted on : Thursday, March 29, 2018 - 11:00:14 AM
Last modification on : Tuesday, January 29, 2019 - 1:21:55 AM

Identifiers

  • HAL Id : tel-01747447, version 1

Collections

Citation

Elyès Alouini. Etude et mise en oeuvre de la réécriture conditionnelle concurrente sur des machines parallèles à mémoire distribuée. Autre [cs.OH]. Université Henri Poincaré - Nancy 1, 1997. Français. ⟨NNT : 1997NAN10011⟩. ⟨tel-01747447⟩

Share

Metrics

Record views

3