Umberto Scafuri

Biological invasion-inspired migration in distributed evolutionary algorithms

Migration strategy plays an important role in designing effective distributed evolutionary algorithms. In this work, a novel migration model inspired to the phenomenon known as biological invasion is devised. The migration strategy is implemented through a multistage process involving invading subpopulations and their competition with native individuals. Such a general approach is used within a stepping-stone parallel model adopting Differential Evolution as the local algorithm. The resulting distributed algorithm is evaluated on a wide set of classical test functions against a large number of sequential and other distributed versions of Differential Evolution available in literature. The findings show that, in most of the cases, the proposed algorithm is able to achieve better performance in terms of both solution quality and convergence rate.

Satellite Image Registration by Distributed Differential Evolution

In this paper a parallel software system based on Differential Evolution for the registration of images is designed, implemented and tested on a set of 2---D remotely sensed images on two problems, i.e. mosaicking and changes in time. Registration is carried out by finding the most suitable affine transformation in terms of maximization of the mutual information between the first image and the transformation of the second one, without any need for setting control points. A coarse---grained distributed version is implemented on a cluster of personal computers.

Extremal Optimization applied to load balancing in execution of distributed programs

The paper describes methods for using Extremal Optimization (EO) for processor load balancing during execution of distributed applications. A load balancing algorithm for clusters of multicore processors is presented and discussed. In this algorithm the EO approach is used to periodically detect the best tasks as candidates for migration and for a guided selection of the best computing nodes to receive the migrating tasks. To decrease the complexity of selection for migration, the embedded EO algorithm assumes a two-step stochastic selection during the solution improvement based on two separate fitness functions. The functions are based on specific models which estimate relations between the programs and the executive hardware. The proposed load balancing algorithm is assessed by experiments with simulated load balancing of distributed program graphs. The algorithm is compared against a greedy fully deterministic approach, a genetic algorithm and an EO-based algorithm with random placement of migrated tasks.

Distributed Differential Evolution for the Registration of Remotely Sensed Images

This paper deals with the design and implementation of a parallel software system based on differential evolution for the registration of images, and with its testing on two bidimensional remotely sensed images on mosaicking problem. Registration is carried out by finding the most suitable affine transformation in terms of maximization of the mutual information between the first image and the transformation of the second one, without any need for setting control points. A coarse-grained distributed version is implemented on a cluster of personal computers

A Distributed Differential Evolution Approach for Mapping in a Grid Environment

Increase in intensive applications with different computational requirements, coupled with the unification of remote and diverse resources thanks to advances in the wide-area network technologies and the low cost of components, have encouraged the development of grid computing. To exploit the promising potentials of geographically distributed resources, effective and efficient mapping algorithms are fundamental. Since the problem of optimally mapping is NP-complete, the development of evolutionary techniques to find near-optimal solutions is welcome. In this paper a distributed system based on differential evolution is designed and implemented to face the mapping problem in a grid environment aiming at reducing the degree of use of the grid resources. This system is tested on some different resource allocation scenarios

An adaptive invasion-based model for distributed Differential Evolution

A novel adaptive model for a recently devised distributed Differential Evolution algorithm is introduced. The distributed algorithm, following the stepping-stone model, is characterized by a migration model inspired by the phenomenon known as biological invasion. The adaptive model is endowed with three updating schemes to randomly set the mutation and the crossover parameters. These schemes are here tied to the migration and are guided by a performance measure between two consecutive migrations. The proposed adaptive model is tested on a set of classical benchmark functions over the different setting schemes. To evaluate its performance, the model is compared against the original non-adaptive version with a fixed parameter setting, and against a well-known distributed Differential Evolution algorithm equipped with the same schemes for the control parameter updating. The experimental study shows that the method results in high effectiveness in terms of solutions detected and convergence speed on most of the benchmark problems and for the majority of the setting schemes investigated. Finally, to further estimate its effectiveness, the proposed approach is also compared with several state-of-the-art Differential Evolution frameworks endowed with different randomized or self-adaptive parameter setting strategies. This comparison shows that our adaptive model allows obtaining the best performance in most of the tests studied.

MGF: A grid-enabled MPI library

Computational grids allow access to several computing resources interconnected in a distributed heterogeneous infrastructure for parallel computing. This powerful resource aggregation increases the application runtime environment complexity. A simple programming model, capable of hiding this complexity, facilitates the use of grid technology in high-performance computing. The message passing interface can play this role and make the grid more accessible to developers with parallel programming skills. In this paper we present MGF, a grid-enabled MPI implementation which extends the existing MPICH-G2. MGF aims are: to allow the transparent use of coupled Grid resources within the MPI library; to give programmers a detailed view of the execution system network topology; to use the most efficient channel available for point-to-point communications and finally, to improve collective operation efficiency by introducing a delegation mechanism.

A Multiobjective Extremal Optimization Algorithm for Efficient Mapping in Grids

Extremal Optimization is proposed to map the tasks making up a user application in grid environments. To comply at the same time with minimal use of grid resources and maximal hardware reliability, a multiobjective version based on the concept of Pareto dominance is developed. The proposed mapper is tested on eight different experiments representing a suitable set of typical real-time situations.

Two new fast heuristics for mapping parallel applications on cloud computing

Abstract In this paper two new heuristics, named Min–min-C and Max–min-C, are proposed able to provide near-optimal solutions to the mapping of parallel applications, modeled as Task Interaction Graphs, on computational clouds. The aim of these heuristics is to determine mapping solutions which allow exploiting at best the available cloud resources to execute such applications concurrently with the other cloud services. Differently from their originating Min–min and Max–min models, the two introduced heuristics take also communications into account. Their effectiveness is assessed on a set of artificial mapping problems differing in applications and in node working conditions. The analysis, carried out also by means of statistical tests, reveals the robustness of the two algorithms proposed in coping with the mapping of small- and medium-sized high performance computing applications on non-dedicated cloud nodes.

Load balancing in distributed applications based on extremal optimization

The paper shows how to use Extremal Optimization in load balancing of distributed applications executed in clusters of multicore processors interconnected by a message passing network. Composed of iterative optimization phases which improve program task placement on processors, the proposed load balancing method discovers dynamically the candidates for migration with the use of an Extremal Optimization algorithm and a special quality model which takes into account the computation and communication parameters of the constituent parallel tasks. Assessed by experiments with simulated load balancing of distributed program graphs, a comparison of the proposed Extremal Optimization approach against a deterministic approach based on a similar load balancing theoretical model is provided.