Danny Munera

A Parametric Framework for Cooperative Parallel Local Search

In this paper we address the problem of parallelizing local search. We propose a general framework where different local search engines cooperate (through communication) in the quest for a solution. Several parameters allow the user to instantiate and customize the framework, like the degree of intensification and diversification. We implemented a prototype in the X10 programming language based on the adaptive search method. We decided to use X10 in order to benefit from its ease of use and the architectural independence from parallel resources which it offers. Initial experiments prove the approach to be successful, as it outperforms previous systems as the number of processes increases.

Solving the Quadratic Assignment Problem with Cooperative Parallel Extremal Optimization

Several real-life applications can be stated in terms of the Quadratic Assignment Problem. Finding an optimal assignment is computationally very difficult, for many useful instances. We address this problem using a local search technique, based on Extremal Optimization and present experimental evidence that this approach is competitive. Moreover, cooperative parallel versions of our solver improve performance so much that large and hard instances can be solved quickly.

Hybridization as Cooperative Parallelism for the Quadratic Assignment Problem

The Quadratic Assignment Problem is at the core of several real-life applications. Finding an optimal assignment is computationally very difficult, for many useful instances. The best results are obtained with hybrid heuristics, which result in complex solvers. We propose an alternate solution where hybridization is obtain by means of parallelism and cooperation between simple single-heuristic solvers. We present experimental evidence that this approach is very efficient and can effectively solve a wide variety of hard problems, often surpassing state-of-the-art systems.

Towards Parallel Constraint-Based Local Search with the X10 Language

In this study, we started to investigate how the Partitioned Global Address Space (PGAS) programming language X10 would suit the implementation of a Constraint-Based Local Search solver. We wanted to code in this language because we expect to gain from its ease of use and independence from specific parallel architectures. We present our implementation strategy, and quest for different sources of parallelism. We discuss the algorithms, their implementations and present a performance evaluation on a representative set of benchmarks.

A framework for simulation and evaluation of cross-layer techniques in wireless sensor networks

Cross-Layer design is a new technique which can be used to improve the performance in wireless sensor networks under the severe restrictions of the energy consumption. Cross-layer techniques intended to create an optimization of the limited resources, taking into account factors associated with different layers of traditional communication schemes, opening a new way for the development of complex communications mechanism. However, because these techniques are based on a non-standard architecture, its structure has a lot of dissimilar characteristics that obstruct a comprehensive comparison of its performance. In this paper, we propose a framework for simulation and evaluation of cross-layer techniques in wireless sensor networks. Our framework promotes a fairness comparison of different communication schemes, a realistic model simulation, an experiment design and a rigorous analysis of results with ANOVA (Analysis of Variance) technique. As a case study, we test our comparative simulation framework with to two popular cross-layer schemes showing the facility of use and the result of the analysis of the data.

Weaving of Metaheuristics with Cooperative Parallelism.

We propose PHYSH (Parallel HYbridization for Simple Heuristics), a framework to ease the design and implementation of hybrid metaheuristics via cooperative parallelism. With this framework, the user only needs encode each of the desired metaheuristics and may rely on PHYSH for parallelization, cooperation and hybridization. PHYSH supports the combination of population-based and single-solution metaheuristics and enables the user to control the tradeoff between intensification and diversification. We also provide an open-source implementation of this framework which we use to model the Quadratic Assignment Problem (QAP) with a hybrid solver, combining three metaheuristics. We present experimental evidence that PHYSH brings significant improvements over competing approaches, as witness the performance on representative hard instances of QAP.

Parallel Local Search

Local Search metaheuristics are a recognized means of solving hard combinatorial problems. Over the last couple of decades, significant advances have been made in terms of the formalization, applicability and performance of these methods. Key to the performance aspect is the increased availability of parallel hardware, which turns out to be largely exploitable by this class of procedures. As the real-life cases of combinatorial optimisation easily degrade into intractable territory for exact or approximation algorithms, local search metaheuristics hold undeniable interest. This situation is further compounded by the good adequacy exhibited by this class of search procedures for large-scale parallel operation. In this chapter we explore and discuss ways which lead to parallelization in Local Search.