Jason Digalakis - Academia.edu (original) (raw)
Uploads
Papers by Jason Digalakis
In this paper we propose a parallel memetic algorithm which combines population-based method with... more In this paper we propose a parallel memetic algorithm which combines population-based method with guided local search (GLS) procedure. In the proposed algorithm, a GLS procedure is applied to each solution generated by genetic operations. The performance of proposed method compared with parallel genetic approaches (i.e. global model, migration model for GA). The parallel implementation is based on message passing interface (MPI).
This paper analyzes some technical and practical issues con-cerning the heterogeneous execution o... more This paper analyzes some technical and practical issues con-cerning the heterogeneous execution of parallel island-based real coded memetic algorithms (PMAs). The parallel environment used is the Mes-sage Passing Interface (MPI). Capability and effectiveness of the PMAs method are discussed fourteen benchmarking functions. The effect of real code representation to the parallel memetic algorithms can make multimodal and non-linear problems easier to solve. In this paper, the characterisitics of the typical two models of parallel memetic algorithms are compared. Those models are the master slave model and the island model. In the master slave model, the ideal parallel efficiency cannot be reached to 100%. Therefore, it is concluded the island model is suitable for PC cluster systems. In addition, we study some special features of the running platforms for PMAs, and basically find out that heterogeneous computing can be as efficient or even more efficient than homogeneous computing for...
The maintenance scheduling problem has been previously tackled by various traditional optimisatio... more The maintenance scheduling problem has been previously tackled by various traditional optimisation techniques. While these methods can give an optimal solution to small scale problems, they are often inefficient when applied to larger scale problems. The memetic algorithm presented here is essentially a genetic algorithm with an element of local search. The effectiveness of the method is tested through its application to real scale problems.
The thermal generator maintenance scheduling problem has been tackled by a variety of traditional... more The thermal generator maintenance scheduling problem has been tackled by a variety of traditional optimisation techniques over the years. This paper proposes a method to solve the maintenance scheduling problem, called parallel co-operating memetic algorithm (PARME). In the proposed model used a variety of selection mechanism, operators, communication methods, local search procedures are applied to each solution generated by genetic operators, and parameters as it explained in the sequel. The PARME alone have been found to produce good quality results. High performance of our approach is demonstrated by applying it to maintenance scheduling problem.
This paper presents experimental results on the major benchmarking functions used for performance... more This paper presents experimental results on the major benchmarking functions used for performance evaluation of Genetic Algorithms (GAs). Parameters considered include the eect of population size, crossover probability, mutation rate and pseudorandom generator. The general computational behavior of two basic GAs models, the Generational Replacement Model (GRM) and the Steady State Replacement Model (SSRM) is evaluated.
The thermal generator maintenance scheduling problem has been tackled by a variety of traditional... more The thermal generator maintenance scheduling problem has been tackled by a variety of traditional optimisation techniques over the years. This paper proposes a method to solve the maintenance scheduling problem, called parallel co-operating memetic algorithm (PARME). In the proposed model used a variety of selection mechanism, operators, communication methods, local search procedures are applied to each solution generated by genetic operators, and parameters as it explained in the sequel. The PARME alone have been found to produce good quality results. High performance of our approach is demonstrated by applying it to maintenance scheduling problem.
this paper we examine 10 di#erent functions in order (a) to test specific parameter of a parallel... more this paper we examine 10 di#erent functions in order (a) to test specific parameter of a parallel execution of memetic algorithms and (b) to evaluate the general computational behavior of MAs. The available theoretical analysis on memetic algorithms does not o#er a tool which could help in a generalized adjustment of control parameters, leaving the choice of the proper operators, parameters and mechanisms to depend on the problem's demands, and the experience and preferences of the researcher
To solve real-world discrete optimization problems approximately metaheuristics such as memetic a... more To solve real-world discrete optimization problems approximately metaheuristics such as memetic algorithms and other evolutionary and local search methods are commonly used. For large instances of these problems or those with a lot of hard constraints even fast heuristics require a considerable amount of computational time. We present PARAMENOAS, an object - oriented memetic algorithms library based on C++ and using the MPI message passing interface. It provides an automatic, transparent way of parallelizing memetic algorithms. The efficient communication in PARAMENOAS is the main reason for its success in several real - world applications.
In this paper we propose a parallel memetic algorithm which combines population-based method with... more In this paper we propose a parallel memetic algorithm which combines population-based method with guided local search (GLS) procedure. In the proposed algorithm, a GLS procedure is applied to each solution generated by genetic operations. The performance of proposed method compared with parallel genetic approaches (i.e. global model, migration model for GA). The parallel implementation is based on message passing interface (MPI).
This paper analyzes some technical and practical issues con-cerning the heterogeneous execution o... more This paper analyzes some technical and practical issues con-cerning the heterogeneous execution of parallel island-based real coded memetic algorithms (PMAs). The parallel environment used is the Mes-sage Passing Interface (MPI). Capability and effectiveness of the PMAs method are discussed fourteen benchmarking functions. The effect of real code representation to the parallel memetic algorithms can make multimodal and non-linear problems easier to solve. In this paper, the characterisitics of the typical two models of parallel memetic algorithms are compared. Those models are the master slave model and the island model. In the master slave model, the ideal parallel efficiency cannot be reached to 100%. Therefore, it is concluded the island model is suitable for PC cluster systems. In addition, we study some special features of the running platforms for PMAs, and basically find out that heterogeneous computing can be as efficient or even more efficient than homogeneous computing for...
The maintenance scheduling problem has been previously tackled by various traditional optimisatio... more The maintenance scheduling problem has been previously tackled by various traditional optimisation techniques. While these methods can give an optimal solution to small scale problems, they are often inefficient when applied to larger scale problems. The memetic algorithm presented here is essentially a genetic algorithm with an element of local search. The effectiveness of the method is tested through its application to real scale problems.
The thermal generator maintenance scheduling problem has been tackled by a variety of traditional... more The thermal generator maintenance scheduling problem has been tackled by a variety of traditional optimisation techniques over the years. This paper proposes a method to solve the maintenance scheduling problem, called parallel co-operating memetic algorithm (PARME). In the proposed model used a variety of selection mechanism, operators, communication methods, local search procedures are applied to each solution generated by genetic operators, and parameters as it explained in the sequel. The PARME alone have been found to produce good quality results. High performance of our approach is demonstrated by applying it to maintenance scheduling problem.
This paper presents experimental results on the major benchmarking functions used for performance... more This paper presents experimental results on the major benchmarking functions used for performance evaluation of Genetic Algorithms (GAs). Parameters considered include the eect of population size, crossover probability, mutation rate and pseudorandom generator. The general computational behavior of two basic GAs models, the Generational Replacement Model (GRM) and the Steady State Replacement Model (SSRM) is evaluated.
The thermal generator maintenance scheduling problem has been tackled by a variety of traditional... more The thermal generator maintenance scheduling problem has been tackled by a variety of traditional optimisation techniques over the years. This paper proposes a method to solve the maintenance scheduling problem, called parallel co-operating memetic algorithm (PARME). In the proposed model used a variety of selection mechanism, operators, communication methods, local search procedures are applied to each solution generated by genetic operators, and parameters as it explained in the sequel. The PARME alone have been found to produce good quality results. High performance of our approach is demonstrated by applying it to maintenance scheduling problem.
this paper we examine 10 di#erent functions in order (a) to test specific parameter of a parallel... more this paper we examine 10 di#erent functions in order (a) to test specific parameter of a parallel execution of memetic algorithms and (b) to evaluate the general computational behavior of MAs. The available theoretical analysis on memetic algorithms does not o#er a tool which could help in a generalized adjustment of control parameters, leaving the choice of the proper operators, parameters and mechanisms to depend on the problem's demands, and the experience and preferences of the researcher
To solve real-world discrete optimization problems approximately metaheuristics such as memetic a... more To solve real-world discrete optimization problems approximately metaheuristics such as memetic algorithms and other evolutionary and local search methods are commonly used. For large instances of these problems or those with a lot of hard constraints even fast heuristics require a considerable amount of computational time. We present PARAMENOAS, an object - oriented memetic algorithms library based on C++ and using the MPI message passing interface. It provides an automatic, transparent way of parallelizing memetic algorithms. The efficient communication in PARAMENOAS is the main reason for its success in several real - world applications.