Parallel Hybrid Metaheuristic Research Articles

Hybrid metaheuristic schemes with different configurations and feedback mechanisms for optimal clustering applications

AbstractThis paper addresses the critical gap in the understanding of the effects of various configurations and feedback mechanisms on the performance of hybrid metaheuristics (HMs) in unsupervised clustering applications. Despite the widespread use of HMs due to their ability to leverage multiple optimization methods, the lack of comprehensive studies on their configuration and feedback mechanisms effects often results in sub-optimal clustering performances and premature convergence. To tackle these issues, we introduce two algorithms for implementing eight distinct HM schemes, focusing on the impacts of parallel and serial processing models along with different feedback mechanisms. Our approach involves selecting candidate metaheuristics based on a mix of evolutionary and swarm-based methods, including the k-means algorithm, to form various HM-based clustering schemes. These schemes were then rigorously evaluated across a range of datasets and feedback mechanisms, further assessing their efficiency in the deployment of smart grid base stations. Performance analysis was based on total fitness evaluations, timing capabilities, and clustering accuracy. The results revealed that parallel HMs with decoupled feedback mechanisms performed best in terms of accuracy but at the cost of slower convergence rates as compared to serial HMs. Our findings further suggest that serial HMs will be best suited for time-sensitive applications where a compromise between speed and accuracy is acceptable, while parallel HMs with decoupled feedback mechanisms are preferable for scenarios where precision is paramount. This research significantly contributes to the field by providing a detailed analysis of HM performance in varying conditions, thereby guiding the selection of appropriate HM schemes for specific clustering tasks.

Pallet location and job scheduling in a Twin-Robot system

This paper introduces the Twin-Robot Pallet Assignment and Scheduling Problem (TRPASP) in which two robots operating on a rail must be scheduled to pick up and deliver a set of products. The objective is to minimise the makespan, defined as the time taken for the robots to transfer all products from their pickup locations to their delivery locations, and return to their starting positions. Pickup locations are known a priori, while delivery locations must be assigned to minimise the makespan. The robots must respect a safety distance to avoid collisions. The paper presents a mathematical model for the TRPASP before introducing four heuristic algorithms for solving the problem. Computational experiments demonstrate that the best results are returned by a parallel hybrid metaheuristic.

Collaborative Parallel Hybrid Metaheuristics on Graphics Processing Unit

Metaheuristics are nondeterministic optimization algorithms used to solve complex problems for which classic approaches are unsuitable. Despite their effectiveness, metaheuristics require considerable computational power and cannot easily be used in time critical applications. Fortunately, those algorithms are intrinsically parallel and have been implemented on shared memory systems and more recently on graphics processing units (GPUs). In this paper, we present highly efficient parallel implementations of the particle swarm optimization (PSO), the genetic algorithm (GA) and the simulated annealing (SA) algorithm on GPU using CUDA. Our approach exploits the parallelism at the solution level, follows an island model and allows for speedup up to 346× for different benchmark functions. Most importantly, we also present a strategy that uses the generalized island model to integrate multiple metaheuristics into a parallel hybrid solution adapted to the GPU. Our proposed solution uses OpenMP to heavily exploit the concurrent kernel execution feature of recent NVIDIA GPUs, allowing for the parallel execution of the different metaheuristics in an asynchronous manner. Asynchronous hybrid metaheuristics has been developed for multicore CPU, but never for GPU. The speedup offered by the GPU is far superior and key to the optimization of solutions to complex engineering problems.

GpuMF: a framework for parallel hybrid metaheuristics on GPU with application to the minimisation of harmonics in multilevel inverters

Metaheuristics are non-deterministic optimisation algorithms used to solve complex problems for which classic approaches are unsuitable or unable to generate satisfying solutions in a reasonable time. Despite their effectiveness, metaheuristics require considerable computational power. Multiple efforts have been made on the development of parallel metaheuristics on graphics processing units (GPUs). Based on a massively parallel architecture, the GPU offers remarkable computing power and can provide significant speedup. However, there currently exists no software project that unites these research initiatives into a comprehensive and reusable tool. To address this shortcoming, we developed gpuMF, a framework for parallel hybrid metaheuristics on GPUs. GPU metaheuristic framework (gpuMF) exploits the intrinsic parallelism found in metaheuristics and fully utilises the massively parallel architecture of GPUs. To demonstrate the effectiveness of our framework, we use gpuMF to minimise the harmonics of multilevel inverters while providing a speedup of 276x.

PARALLEL HYBRID METAHEURISTIC ON SHARED MEMORY SYSTEM FOR REAL-TIME UAV PATH PLANNING

In this paper, we present a parallel hybrid metaheuristic that combines the strengths of the particle swarm optimization (PSO) and the genetic algorithm (GA) to produce an improved path-planner algorithm for fixed wing unmanned aerial vehicles (UAVs). The proposed solution uses a multi-objective cost function we developed and generates in real-time feasible and quasi-optimal trajectories in complex 3D environments. Our parallel hybrid algorithm simulates multiple GA populations and PSO swarms in parallel while allowing migration of solutions. This collaboration between the GA and the PSO leads to an algorithm that exhibits the strengths of both optimization methods and produces superior solutions. Moreover, by using the "single-program, multiple-data" parallel programming paradigm, we maximize the use of today's multicore CPU and significantly reduce the execution time of the parallel program compared to a sequential implementation. We observed a quasi-linear speedup of 10.7 times faster on a 12-core shared memory system resulting in an execution time of 5 s which allows in-flight planning. Finally, we show with statistical significance that our parallel hybrid algorithm produces superior trajectories to the parallel GA or the parallel PSO we previously developed.

A PARALLEL METAHEURISTIC FRAMEWORK BASED ON HARMONY SEARCH FOR SCHEDULING IN DISTRIBUTED COMPUTING SYSTEMS

A large number of optimization problems have been identified as computationally challenging and/or intractable to solve within a reasonable amount of time. Due to the NP-hard nature of these problems, in practice, heuristics account for the majority of existing algorithms. Metaheuristics are one very popular type of heuristics used for many of these optimization problems. In this paper, we present a novel parallel-metaheuristic framework, which effectively enables to devise parallel metaheuristics, particularly with heterogeneous metaheuristics. The core component of the proposed framework is its harmony-search-based coordinator. Harmony search is a recent breed of metaheuristic that mimics the improvisation process of musicians. The coordinator facilitates heterogeneous metaheuristics (forming a parallel metaheuristic) to escape local optima. Specifically, best solutions generated by these worker metaheuristics are maintained in the harmony memory of the coordinator, and they are used to form new-possibly better-harmonies (solutions) before actual solution sharing between workers occurs; hence, their solutions are harmonized with each other. For the applicability validation and the performance evaluation, we have implemented a parallel hybrid metaheuristic using the framework for the task scheduling problem on multiprocessor computing systems (e.g., computer clusters). Experimental results verify that the proposed framework is a compelling approach to parallelize heterogeneous metaheuristics.

Parallel hybrid metaheuristics for the flexible job shop problem

A parallel approach to flexible job shop scheduling problem is presented in this paper. We propose two double-level parallel metaheuristic algorithms based on the new method of the neighborhood determination. Algorithms proposed here include two major modules: the machine selection module refer to executed sequentially, and the operation scheduling module executed in parallel. On each level a metaheuristic algorithm is used, therefore we call this method meta 2heuristics. We carry out a computational experiment using Graphics Processing Units (GPU). It was possible to obtain the new best known solutions for the benchmark instances from the literature.

Some aspects concerning the implementation of a parallel hybrid metaheuristic

This paper presents a discrete optimization system implementing a parallel hybrid metaheuristic. This is obtained by joining a genetic algorithm and a parallel version of a stochastic descent method called ‘Kangaroo’. Two real problems in the manufacturing field were solved using the proposed metaheuristic. This offered the opportunity to underline some aspects regarding the implementation of this hybrid system. The impact of the precedence constraints upon the implementation of the genetic operators (crossover and mutation) is also considered.

Designing cellular networks using a parallel hybrid metaheuristic on the computational grid

Cellular network design is a major issue in mobile telecommunication systems. In this paper, a model of the problem in its full practical complexity, based on multiobjective constrained combinatorial optimization, has been investigated. We adopted the Pareto approach at resolution in order to compute a set of diversified non-dominated networks, thus removing the need for the designer to rank or weight objectives a priori. We designed and implemented a “ready-to-use” platform for radio network optimization that is flexible regarding both the modeling of the problem (adding, removing, updating new antagonist objectives and constraints) and the solution methods. It extends the “white-box” ParadisEO framework for metaheuristics applied to the resolution of mono/multi-objective Combinatorial Optimization Problems requiring both the use of advanced optimization methods and the exploitation of large-scale parallel and distributed environments. Specific coding scheme and genetic and neighborhood operators have been designed and embedded. On the other side, we make use of many generic features related to advanced intensification and diversification search techniques, hybridization of metaheuristics and grid computing for the distribution of the computations. They aim at improving the quality of networks and their robustness. They also allow, to speed-up the search and obtain results in a tractable time, and so efficiently solving large instances of the problem. Using three realistic benchmarks, the computed networks and speed-ups on different parallel and/or distributed architectures show the efficiency and the scalability of hierarchical parallel hybrid models.

Grid computing for parallel bioinspired algorithms

This paper focuses on solving large size combinatorial optimization problems using a Grid-enabled framework called ParadisEO–CMW ( Para ̲ llel and Dis ̲ tributed EO ̲ on top on C ̲ ondor and the M ̲ aster W ̲ orker Framework). The latter is an extension of ParadisEO, an open source framework originally intended to the design and deployment of parallel hybrid meta-heuristics on dedicated clusters and networks of workstations. Relying on the Condor–MW framework, it enables the execution of these applications on volatile heterogeneous computational pools of resources. The motivations, architecture and main features will be discussed. The framework has been experimented on a real-world problem: feature selection in near-infrared spectroscopic data mining. It has been solved by deploying a multi-level parallel model of evolutionary algorithms. Experimentations have been carried out on more than 100 PCs originally intended for education. The obtained results are convincing, both in terms of flexibility and easiness at implementation, and in terms of efficiency, quality and robustness of the provided solutions at run time.