Parallel Genetic Algorithm Research Articles

Numerical mode decomposition for multimode fiber: From multi-variable optimization to deep learning

Abstract Multimode fibers are gaining a resurgence of interest in both fundamental and applied research in recent years. Thanks to the ability to see the weights and relative phase of the multimode fiber modes, mode decomposition (MD) has shown tremendous potential in wide applications of mode properties evaluation, mode-related processes measurement and fiber laser beams characterization. Among various MD techniques, numerical methods stand out for their simplicity and low hardware requirements. In this paper, the new horizon opened by the recently developed new numerical MD schemes will be reviewed. First, the background and basic principles of MD will be introduced, and some typical numerical MD methods will be summarized. Second, the multi-variable optimization approaches, including the stochastic parallel gradient descent (SPGD) scheme and genetic algorithm (GA) assisted GA-SPGD strategy, will be presented in details. Third, a novel numerical MD method based on deep learning technique will be discussed, which solves the initial values sensitivity and relatively long-time cost of multi-variable optimization approaches. Last, several novel applications will be given, indicating the versatile applicability of numerical MD.

Application of an Adaptive Multi-Population Parallel Genetic Algorithm with Constraints in Electromagnetic Tomography with Incomplete Projections

Electromagnetic tomography technology (EMT) is widely used in underground energy exploration. Limited by objective conditions, the collected projection data of electromagnetic waves are sparse and incomplete. Therefore, a study of the tomographic inversion algorithm of EMT based on incomplete projection data has an important guiding significance for the exploitation of underground energy. As a global optimization probability search algorithm, the simple genetic algorithm (SGA) has been widely used in the process of tomographic inversion. However, SGA evolves through a single population, and the values of crossover and mutation probability are always fixed, so there are risks of premature convergence and poor local search ability. To improve the performance of the SGA, a new approach of adaptive multi-population parallel genetic algorithm (AMPGA) with constraints is proposed in this paper. First, the AMPGA makes full use of multi-group adaptive co-evolution to improve the local and global search ability of SGA and restrain the risk of premature convergence. Then, the introduction of prior information as a constraint makes the results clearer and more accurate. The proposed algorithm has been verified in a numerical experiment and field tests, and the results show that the proposed algorithm can well balance global and local search capabilities, which offers a more realistic and stable tomographic result.

Performance analysis of synchronous and asynchronous distributed genetic algorithms on multiprocessors

Abstract Because of their effectiveness and flexibility in finding useful solutions, Genetic Algorithms (GAs) are very popular search techniques for solving complex optimization problems in scientific and industrial fields. Parallel GAs (PGAs), and especially distributed ones have been usually presented as the way to overcome the time-consuming shortcoming of sequential GAs. In the case of applying PGAs, we can expect better performance, the reason being the exchange of knowledge during the parallel search process. The resulting distributed search is different compared to what sequential panmictic GAs do, then deserving additional studies. This article presents a performance study of three different PGAs. Moreover, we investigate the effect of synchronizing communications over modern shared-memory multiprocessors. We consider the master-slave model along with synchronous and asynchronous distributed GAs (dGAs), presenting their different designs and expected similarities when running in a number of cores ranging from one to 32 cores. The master-slave model showed a competitive numerical effort versus the other dGAs and demonstrated to be able to scale-up well over multiprocessors. We describe how the speed-up and parallel performance of the dGAs is changing as the number of cores enlarges. Results of the island model show that synchronous and asynchronous dGAs have different numerical performances on a multiprocessor, the asynchronous algorithm having a faster execution, thus more attractive for time demanding applications. Our results and statistical analyses help in developing a novel body of knowledge on PGAs running in shared memory multiprocessors (versus overwhelming literature oriented to distributed memory clusters), something useful for researchers, beginners, and final users of these techniques.

Spark-Based Parallel Genetic Algorithm for Simulating a Solution of Optimal Deployment of an Underwater Sensor Network.

Underwater sensor networks have wide application prospects, but the large-scale sensing node deployment is severely hindered by problems like energy constraints, long delays, local disconnections, and heavy energy consumption. These problems can be solved effectively by optimizing sensing node deployment with a genetic algorithm. However, the genetic algorithm (GA) needs many iterations in solving the best location of underwater sensor deployment, which results in long running time delays and limited practical application when dealing with large-scale data. The classical parallel framework Hadoop can improve the GA running efficiency to some extent while the state-of-the-art parallel framework Spark can release much more parallel potential of GA by realizing parallel crossover, mutation, and other operations on each computing node. Giving full allowance for the working environment of the underwater sensor network and the characteristics of sensors, this paper proposes a Spark-based parallel GA to calculate the extremum of the Shubert multi-peak function, through which the optimal deployment of the underwater sensor network can be obtained. Experimental results show that while faced with a large-scale underwater sensor network, compared with single node and Hadoop framework, the Spark-based implementation not only significantly reduces the running time but also effectively avoids the problem of premature convergence because of its powerful randomness.

Soft Sensing for Gas-Condensate Field Production Using Parallel-Genetic-Algorithm-Based Data Reconciliation

During present offshore gas-condensate production, multiphase flow-meters, due to its exceedingly high cost, are being substituted by a soft sensing (SS) technique for estimating total and single-well flowrates through sensor measurements and physical models. In this work, the inverse problem is solved by data reconciliation (DR), minimizing weighted sum of errors with constraints integrating multiple two-phase flow models. The DR problem is solved by parallel genetic algorithm (PGA) without complex calculations required by conventional optimization. The newly developed SS method is tested by data from a realistic gas-condensate production system. The method is proved of good accuracy and robustness with invalid individual pressure sensor or unavailable total flowrate measurements. Meanwhile, the proposed method shows good parallel performance and the time cost of each DR process can meet the demand of engineering application.

Unraveling Thermodynamic Stability, Catalytic Activity, and Electronic Structure of [TMxMgyOz]+/0/– Clusters at Realistic Conditions: A Hybrid DFT and ab Initio Thermodynamics Study

Aiming toward catalytic applications, a large data set is generated on [TMxMgyOz]+/0/– clusters (TM = Cr, Fe, Co, Ni, x + y ≤ 5) using a massively parallel cascade genetic algorithm (cGA) approach at the hybrid density functional level of theory. The low-energy isomers are further analyzed via ab initio atomistic thermodynamics to estimate their free energy of formation at a realistic temperature T and partial pressure of oxygen pO2. A thermodynamic phase diagram is drawn by minimizing Gibbs free energy of formation to identify the stable phases of neutral and charged [TMxMgyOz]+/0/– clusters. From this analysis, we notice that neutral and negatively charged clusters are stable in the wide range of (T, pO2). The negatively charged clusters are more effective as a catalyst to lower the C–H bond activation barrier for oxidation of methane. We find that the nature of TM atoms toward controlling the activation barrier is less important. However, the TM gives rise to different structural motifs in the cluster, which may act as active centers for catalysis.

Parallel Hierarchical Genetic Algorithm for Scattered Data Fitting through B-Splines

Curve fitting to unorganized data points is a very challenging problem that arises in a wide variety of scientific and engineering applications. Given a set of scattered and noisy data points, the goal is to construct a curve that corresponds to the best estimate of the unknown underlying relationship between two variables. Although many papers have addressed the problem, this remains very challenging. In this paper we propose to solve the curve fitting problem to noisy scattered data using a parallel hierarchical genetic algorithm and B-splines. We use a novel hierarchical structure to represent both the model structure and the model parameters. The best B-spline model is searched using bi-objective fitness function. As a result, our method determines the number and locations of the knots, and the B-spline coefficients simultaneously and automatically. In addition, to accelerate the estimation of B-spline parameters the algorithm is implemented with two levels of parallelism, taking advantages of the new hardware platforms. Finally, to validate our approach, we fitted curves from scattered noisy points and results were compared through numerical simulations with several methods, which are widely used in fitting tasks. Results show a better performance on the reference methods.

Research on heterogeneous distributed fractal image compression algorithm based on parallel genetic algorithm and iterative function

This paper briefly introduces the method and implementation of image compression of heterogeneous distributed genetic and iterative function system (IFS). Based on the traditional parallel genetic algorithm, a heterogeneous distributed model of parallel fractal image compression based on genetic algorithm is implemented, and the corresponding complex model design scheme is put forward. The results show that the algorithm has better search ability, its comprehensive performance such as signal-to-noise ratio/compression time is further improved, the decoding image is similar to the original image, and has high image quality effect.

Optimal Sensors/Actuators Placement in Smart Structure Using Island Model Parallel Genetic Algorithm

Determination of optimal placements of sensors/actuators in large structures is a difficult job as large number of possible combinations leads to a very high computational time and storage. Therefore, this kind of optimization problem demands a parallel implementation of the optimization schemes. Island model genetic algorithm (GA) being inherently parallel has been used for searching optimal placements of collocated sensors/actuators. Numerical simulations have been done for determination of optimal placements of collocated PZT sensors and actuators in smart fiber reinforced shell structures using island model parallel GA (IMPGA) in conjunction with electro-mechanical finite element analysis with an objective of maximizing the controllability index. It has been observed that the present IMPGA-based formulation (due to its migration scheme) not only makes it possible to determine optimal sensors/actuators locations for large structures but also leads to a better solution at a much reduced and achievable computational time. Results from scalability analysis also show that the efficacy of the present method of using IMPGA for determination of optimal sensors/actuators location based on FEA will be more pronounced when actually used for real life problems requiring large number of sensors and actuators.

Throughput optimisation in a coal export system with multiple terminals and shared resources

This work describes a genetic algorithm based approach for the optimization of the Hunter Valley coal export system in Newcastle, Australia. The Port of Newcastle features three coal export terminals, operating primarily in cargo assembly mode. They share a rail network on their inbound operations and a channel on their outbound operations. Maximizing throughput at a single coal terminal, taking into account its layout, equipment and operating policies, is already a challenging problem. However, maximizing throughput of the Hunter Valley coal export system as a whole requires that terminals and inbound/outbound shared resources be considered simultaneously. Existing approaches to solve this and similar problems either lack realism or are computationally too demanding to be useful as an everyday planning tool. We present a parallel genetic algorithm to optimize the integrated system. The algorithm models activities in continuous time and can handle practical planning horizons efficiently. The solutions are on average 17% better than those obtained with the current state-of-the-art method – a constraint programming-based approach – requiring less than 3% of the CPU time. Tests were conducted on 10 instances generated using real world data, with 200 vessels and approximately 270 stockpiles each.

Genetic Feature Construction: a parallel implementation of a genetic programming tool for feature construction


 In this paper a parallel implementation of a recently introduced method for feature construction is described. This technique utilizes parallel genetic algorithms along with RBF neural networks to create new features from the original ones by discovering the hidden relations between patterns. The method is tested on series of classification problems from a variety of areas and the results are reported. The accompanied software is written entirely in ANSI C++ using the well established MPI library for parallelization.

Genetic Feature Construction Genetic Feature Construction: a parallel implementation of a genetic programming tool for feature construction


 In this paper a parallel implementation of a recently introduced method for feature construction is described. This technique utilizes parallel genetic algorithms along with RBF neural networks to create new features from the original ones by discovering the hidden relations between patterns. The method is tested on series of classification problems from a variety of areas and the results are reported. The accompanied software is written entirely in ANSI C++ using the well established MPI library for parallelization.

Hybrid dual-objective parallel genetic algorithm for heterogeneous multiprocessor scheduling

Scheduling is a process of mapping resources to tasks and it’s objective is either one or more. This paper focuses on scheduling in heterogeneous multiprocessor systems. Here the resources are processing elements and tasks are the jobs submitted to the processor. The main objectives of multiprocessor scheduling are reducing schedule length, reducing the overall energy consumption, reducing the temperature, reducing failure rates and so on. A Hybrid dual-objective parallel genetic algorithm is applied in the proposed work. Makespan and energy consumption are the two objectives considered. The proposed algorithm determines the global optimal solutions by generating the initial population using some heuristics and then performing parallel genetic operations on it. The main aim of employing parallelism is to find a global optimum solution by avoiding premature convergence in a local optimum and to reduce the running time of the algorithm. Hill climbing is also used in addition, to avoid local optimum solutions. The proposed algorithm balances the tradeoff between energy consumption and makespan according to the inclinations of the users by following weighted sum methodology. Our experimental results demonstrate that the proposed algorithm outperforms the other existing algorithms in terms of both makespan and energy consumption by incurring less running time.

Estimation of frequency domain soil parameters of horizontally multilayered earth by using Cole–Cole model based on the parallel genetic algorithm

A method to estimate frequency domain soil parameters of horizontally multilayered earth is developed. Cole–Cole model of complex conductivity form is adopted to describe the frequency dependence of soil parameters. The estimation of frequency domain soil parameters includes two stages. In the first stage, soil-layered structure is determined under the DC field by the interpretation of resistivity sounding data. The obtained model parameters in this stage include the DC soil conductivity and thickness of soil layers. In the second stage, Cole–Cole model parameters are estimated based on the soil-layered model obtained in the first stage. The theoretical formula of complex apparent resistivity is derived from Green's function with considering dynamic-state field theory. Since dynamic-state field is considered in the inversion algorithm, this method can be extended into high-frequency domain. Discrete complex image method is used to calculate Sommerfeld integral quickly. In order to remove electromagnetic (EM) coupling between cables at high frequencies, a new electrode configuration method is proposed. Genetic algorithm is improved by combining with parallel computing. The parallel genetic algorithm is applied to the optimisation of model parameters.

A component-based study of energy consumption for sequential and parallel genetic algorithms

Recently, energy efficiency has gained attention from researchers interested in optimizing computing resources. Solving real-world problems using optimization techniques (such as metaheuristics) requires a large number of computing resources and time, consuming an enormous amount of energy. However, only a few and limited research efforts in studying the energy consumption of metaheuristics can be found in the existing literature. In particular, genetic algorithms (GAs) are being used so widely to solve a large range of problems in scientific and real-world problems, but hardly found explained in their internal consumption behavior. In the present article, we analyze the energy consumption behavior of such techniques to offer a useful set of findings to researchers in the mentioned domains. We expand our study to include several algorithms and different problems and target the components of the algorithms so that the results are still more appealing for researchers in arbitrary domains of application. Our experiments on the sequential GAs show the controlling role of the fitness operator on energy consumption and also reveal possible energy hot spots in GAs operations, such as mutation operator. Further, our distributed evaluations besides a statistical analysis of the results demonstrate that the communication scheme could highly affect the energy consumption of the parallel evaluations of the GAs.

Developing parallel real-coded genetic algorithms for decision-making systems of socio-ecological and economic planning

Developing parallel real-coded genetic algorithms for decision-making systems of socio-ecological and economic planning

Uncertainty analysis of water distribution networks using type-2 fuzzy sets and parallel genetic algorithm

ABSTRACTWater distribution networks are typically designed based on certain values of uncertain parameters such as reservoir head, nodal demand and friction coefficient to furnish acceptable performance by limiting nodal pressure head and pipes’ velocities. In this study, a fuzzy uncertainty analysis approach is developed to handle uncertainty of reservoir head and nodal demand over an extended period and long-term simulation, based on type-2 fuzzy logic. Furthermore, the long-term uncertainty of Hazen-Williams (HW) coefficient is also considered. A parallel genetic algorithm is introduced to solve a multi-objective optimization problem for fuzzy uncertainty analysis. The method is powered by a pressure dependent hydraulic simulator which is based on EPANET and an iterative procedure. It is implemented for a benchmark network from literature and a case study. Results show that network velocity is highly affected by uncertainties and the accumulation of different uncertainties may change the network performance significantly.

Parallel multi-agent real-coded genetic algorithm for large-scale black-box single-objective optimisation

There is an ongoing evolution involving a new approach to large-scale optimisations based on co-evolutionary searches using interacting heterogeneous agent-processes via the implementation of synchronised genetic algorithms with local populations. The individualisation of heuristic operators at the level of agent-processes that implement independent evolutionary searches facilitate the improved likelihood of obtaining the best solutions in the fastest time. Based on this property, a parallel multi-agent single-objective real-coded genetic algorithm for large-scale constrained black-box single-objective optimisations (LSOPs) is proposed. This facilitates the effective frequency exchange of the best potential decisions between interacting agent-processes with individual parameters, such as types of crossover and mutation operators with their own characteristics. We have improved the quality of both solutions and the time-efficiency of a multi-agent real-coded genetic algorithm (MA−RCGA). A novel framework was developed that represents the aggregation of MA−RCGA with simulation models by implementing a set of objective functions for real-world large-scale optimisation problems such as the simulation model of the ecological-economics system implemented in the AnyLogic tool.

A parallel multi-objective genetic algorithm with learning based mutation for railway scheduling

Railway system is a reliable and efficiency major public transportation. It is supported by many countries since it has a less environmental effect compared to another type of transportation. As the railway networks have become larger and more complex with increasing passenger demand, both aspects from the passenger satisfaction and operational cost need to be satisfied. This paper proposes a Parallel Multi-objective Evolutionary Algorithm with Hybrid Sampling Strategy and learning-based mutation to solve the railway train scheduling problem. Learning techniques have been coupled with a multi-objective genetic algorithm to guide the search for better solutions. In this paper, we incorporate a learning-based algorithm into a mutation process. The evaluation process is divided into sub-process and calculated by a parallel computational unit using GPU CUDA framework. Two sets of numerical experiments based on a small-scale case of Thailand ARL transit line and a larger case of BTS transit network are implemented to verify the effectiveness of the proposed approaches. The experimental results show the effectiveness of the proposed algorithm comparing to sequential CPU computational and two classical multi-objective evolutionary algorithms. With the same number of operating trains, the proposed algorithm can obtain schedule with less average waiting time and the time used for computational is significantly reduced.

Design and energy management optimization for hybrid renewable energy system- case study: Laayoune region

The major objectives of this work are: 1) to develop new efficient optimization algorithm to solve NP-hard problems, 2) to show the potential of integrating renewable energy technologies for Laayoune region- Morocco taken as case study. In this work, new parallel hybrid Genetic Algorithm-Particle Swarm Optimization algorithm (P-GA-PSO) is developed to solve both sizing and energy management problems for micro-grids. The studied micro-grid is composed of four different renewable energy technologies and energy storage system. The objectives of both optimization problems are to satisfy typical load demand, to minimize energy production cost, to maximize renewable energy integration, to avoid energy losses and overload. P-GA-PSO performances are evaluated and compared with the ordinary optimization methods based on: 1) a set of benchmark functions, 2) various scenarios used to solve the proposed micro-grid optimization problems. Obtained results demonstrated that P-GA-PSO outperforms ordinary optimization algorithms in terms of convergence time and solution quality. Moreover, the proposed micro-grid is very promising in terms of load demand satisfaction, cost and pollutant emissions reduction. Indeed, the obtained energy cost does not exceed 0.17 US$/kWh, which is close to fossil fuel energy cost, and fossil fuel replacement rate exceeds 50% during all periods.