Ewa Szlachcic

Fault Tolerant Topological Design for Computer Networks

The fault-tolerant topological design for a computer network leads one to characterize the way in which the nodes are linked to each other with the known connectivity parameter and to the capacity of links, which represent the means of transmission parameters between the vertices. The design problem is to find a suitable fault tolerant network topology at a minimum communication cost under the constraint of an average packet time delay. An approach based on evolutionary algorithm (EA) is developed for the network topological design problem. The special construction of a chromosome according to fault tolerant network configuration was designed and the modification of fitness function is proposed. Simulations are studied to support the effectiveness of the proposed algorithm

Interactive procedures in large-scale two-dimensional cutting stock problems

Abstract This paper discusses basic problem formulation, and solution procedures for solving a particular large-scale two-dimensional cutting stock problem at a furniture factory. In recent years there has been an explosion of interests in this application area. The cutting stock problems are met in many branches of industry. There is a large economical incentive to find more effective solution procedures, and it is easy to compare alternative solution procedures and to identify potential benefits of using a proposed procedure. The paper concentrates on the case in which both stock and ordered sizes are rectangular. We present a mathematical model of the problem, propose an optimization algorithm and describe its basic properties. The mathematical model of the optimization problem relies on the Gilmore and Gomory approach. To decrease the total computing time we propose an interactive procedure based on the forecasting of the criterion functions values.

Differential evolution for multi-objective optimization with self adaptation

In the paper an adapted version of the differential evolution algorithm has been created to solve a multi-objective optimization problem. Multi-objective Differential Evolution Algorithm using vector differences for perturbing the vector population with self adaptation is introduced. Through the combination of mutation strategies and self adaptation of crossover and differentiation constants the proposed MO algorithm performs better than the one with the simple DE scheme in terms of computation speed and quality of the generated multi-objective non-dominated solutions.

Efficiency Analysis in Communication Networks Topology Design

Increasing attention is being recently devoted to various problems in the topological design of communication networks. In the paper we propose to solve a bi-criteria network topology design problem for considering a message delay and global cost as an objective functions vector under the connectivity constraint. An approach based on the bi-criteria genetic algorithm VEGA is developed for the network topology design problem. We present some experiments in order to certify the influence of genetic algorithm parameters for the quality of optimal design variable space and of an objective functions space. Finally a global network efficiency measure is proposed as an indicator for the designer to determine an efficient network communication topology from the Pareto-optimal set of solutions. Numerical results provide to illustrate that the proposed methodology can search effectively one communication network topology

Decision Support System for Cancer Chemotherapy Schedules

In a chemotherapy scheduling process a chemotherapy is a treatment of cancer using a set of toxic drugs. In the paper we propose a Decision Support System for the anti-cancer medical treatment to improve physicians’ decisions about drugs doses selection and scheduling. A hybrid meta-heuristic algorithm has been applied to the problem of bi-criteria optimization allowing to find effective chemotherapy drugs dose scheduling as the minimization of a tumor size at a fixed period of time and maximization of Patient Survival Time. The numerical tests of proposed algorithm gives the possibility of producing a set of alternative treatment scenarios according to the final decision.

The cutting stock problem recent application in information technology

Cloud computing is a Web-based processing, that allow to share resources, software, and information over the Internet. Cloud computing helps enterprises and other institutions like schools, universities, etc. transform business and technology. Most cloud computing infrastructures consist of services delivered through common centers and built on servers.

Parallel Distributed Genetic Algorithm for Expensive Multi-Objective Optimization Problems

In many Multi-Objective Optimization Problems it is required to evaluate a great number of objective functions and constraints and the calculation effort is very high. The use of parallelism in Multi-Objective Genetic Algorithms is one of the solutions of this problem. In this work we propose an algorithm, based on parallelization scheme using island model with spatially isolated populations. The intent of the proposed paper is to illustrate that modifications made to a selection and resolution processes and to a migration scheme have further improved the efficiency of the algorithm and good distribution of Pareto front.

An Adaptive Selection Evolutionary Algorithm for the Capacitated Vehicle Routing Problem

We propose a meta-heuristic based on an evolutionary approach for a Capacitated Vehicle Routing Problem. The modifications concern a selection process and two new heuristics for crossover operators. The numerical results demonstrate the effectiveness of an adaptive selection evolutionary algorithm on the benchmark test problems. The main advantage is the possibility of arranging the proposed selection process and crossover operators in the space of feasible solutions. The presented results are very promising for solving bigger problems. number of customers is large. The largest problems which can be consistently solved by the most effective exact algorithms proposed so far contain about 50 customers, whereas larger instances may be solved only in particular cases. So instances with hundreds of customers, as those arising in practical applications, may only be tackled with heuristic or meta- heuristic methods (12). In meta-heuristics, the emphasis is on performing a deep exploration of the most promising regions of the solution space. The quality of solutions produced by these methods is much better than that obtained by classical heuristic methods. In the paper special attention is paid to meta-heuristics methods with evolutionary mechanisms based on the idea of natural search in biology (5,8,12). Our purpose is to propose an evolutionary search based meta-heuristic idea with modified tournament selection process for CVRP. We will describe an adaptive tournament selection and two new crossover operators for the discussed routing problem.

Differential Evolution Multi-objective Optimisation for Chemotherapy Treatment Planning

Differential evolution is currently one of the most popular population based stochastic meta-heuristics. In the paper, we propose an extension of the Differential Evolution algorithm for multi-objective optimization problem with constraints of chemotherapy scheduling for a medical treatment. The differential evolution idea is used with some significant improvements concerning the DE strategies and parameters adaptation. The numerical results show that the proposed algorithm is stable and robust in handling medical applications especially for a chemotherapy planning process.

Some numerical aspects of the gradient procedure for design of nonlinear transportation networks

In the paper a few algorithms for determination the parameters of nonlinear transportation networks with heuristic starting points procedures are analyzed. The suitable illustrations of the proposed algorithms are based on the water distribution networks and computer networks.