A hybrid optimization technique coupling an evolutionary and a local search algorithm

Evolutionary algorithms are robust and powerful global optimization techniques for solving large-scale problems that have many local optima. However, they require high CPU times, and they are very poor in terms of convergence performance. On the other hand, local search algorithms can converge in a few iterations but lack a global perspective. The combination of global and local search procedures should offer the advantages of both optimization methods while offsetting their disadvantages. This paper proposes a new hybrid optimization technique that merges a genetic algorithm with a local search strategy based on the interior point method. The efficiency of this hybrid approach is demonstrated by solving a constrained multi-objective mathematical test-case.     

A new heuristic algorithm solving the linear ordering problem

The linear ordering problem is an NP-hard combinatorial problem with a large number of applications. Contrary to another very popular problem from the same category, the traveling salesman problem, relatively little space in the literature has been devoted to the linear ordering problem so far. This is particularly true for the question of developing good heuristic algorithms solving this problem.In the paper we propose a new heuristic algorithm solving the linear ordering problem. In this algorithm we made use of the sorting through insertion pattern as well as of the operation of permutation reversal. The surprisingly positive effect of the reversal operation, justified in part theoretically and confirmed in computational examples, seems to be the result of a unique property of the problem, called in the paper the symmetry of the linear ordering problem. This property consists in the fact that if a given permutation is an optimal solution of the problem with the criterion function being maximized, then the reversed permutation is a solution of the problem with the same criterion function being minimized.     

Genetic Algorithm for the Jump Number Scheduling Problem

This paper introduces genetic algorithms for the jump number scheduling problem. Given a set of tasks subject to precedence constraints, the problem is to construct a schedule to minimize the number of jumps. We show that genetic algorithms outperform the previously known Knuth and Szwarcfiter's exhaustive search algorithm when applied to some classes of orders in which no polynomial time algorithms exist in solving the jump number problem. Values for various parameters of genetic jump number algorithms are tested and results are discussed.     

A new branch and bound algorithm for cell formation problem

Cell formation (CF) is the first and the most important problem in designing cellular manufacturing systems. Due to its non-polynomial nature, various heuristic and metaheuristic algorithms have been proposed to solve CF problem. Despite the popularity of heuristic algorithms, few studies have attempted to develop exact algorithms, such as branch and bound (B&B) algorithms, for this problem. We develop three types of branch and bound algorithms to deal with the cell formation problem. The first algorithm uses a binary branching scheme based on the definitions provided for the decision variables. Unlike the first algorithm, which relies on the mathematical model, the second one is designed based on the structure of the cell formation problem. The last algorithm has a similar structure to the second one, except that it has the ability to eliminate duplicated nodes in branching trees. The proposed branch and bound algorithms and a hybrid genetic algorithm are compared through some numerical examples. The results demonstrate the effectiveness of the modified problem-oriented branch and bound algorithm in solving relatively large size cell formation problems.     

Successive column correction algorithms for solving sparse nonlinear systems of equations

This paper presents two algorithms for solving sparse nonlinear systems of equations: the CM-successive column correction algorithm and a modified CM-successive column correction algorithm. Aq-superlinear convergence theorem and anr-convergence order estimate are given for both algorithms. Some numerical results and the detailed comparisons with some previously established algorithms show that the new algorithms have some promise of being very effective in practice.This research was partially supported by contracts and grants: DOE DE-AS05-82ER1-13016, AFOSR 85-0243 at Rice University, Houston, U.S.A. and Natural Sciences and Engineering Research Council of Canada grant A-8639.     

Integer extended ABS algorithms and possible control of intermediate results for linear Diophantine systems

We present a new class of integer extended ABS algorithms for solving linear Diophantine systems. The proposed class contains the integer ABS (the so-called EMAS and our proposed MEMAS) algorithms and the generalized Rosser’s algorithm as its members. After an application of each member of the class a particular solution of the system and an integer basis for the null space of the coefficient matrix are at hand. We show that effective algorithms exist within this class by appropriately setting the parameters of the members of the new class to control the growth of intermediate results. Finally, we propose two effective heuristic rules for selecting certain parameters in the new class of integer extended ABS algorithms.      

Bees algorithm for generalized assignment problem

Bees algorithm (BA) is a new member of meta-heuristics. BA tries to model natural behavior of honey bees in food foraging. Honey bees use several mechanisms like waggle dance to optimally locate food sources and to search new ones. This makes them a good candidate for developing new algorithms for solving optimization problems. In this paper a brief review of BA is first given, afterwards development of a BA for solving generalized assignment problems (GAP) with an ejection chain neighborhood mechanism is presented. GAP is a NP-hard problem. Many meta-heuristic algorithms were proposed for its solution. So far BA is generally applied to continuous optimization. In order to investigate the performance of BA on a complex integer optimization problem, an attempt is made in this paper. An extensive computational study is carried out and the results are compared with several algorithms from the literature.     

On solving the forward kinematics of 3RPR planar parallel manipulator using hybrid metaheuristics

Hybrid metaheuristics have been applied with success in solving many real-world problems. This work introduces hybrid metaheuristics to the field of kinematics problem, in particular, for solving the forward kinematics of the 3RPR parallel manipulator. It implements a combination of genetic algorithms and simulated annealing into two popular hybrid metaheuristic techniques. They are combined as teamwork and relay collaborative hybrid metaheuristics and compared to the performance of genetic algorithms and simulated annealing alone. The results show that the meta-heuristic approaches give robust and high quality solutions. Genetic algorithms and teamwork collaborative metaheuristics showed better performance than simulated annealing and relay collaborative metaheuristics. The given metaheuristic methods obtain all the unique solutions and comparisons with algebraic methods show promising results.     

Distributed computation of Pareto solutions inn-player games

The problem of computing Pareto optimal solutions with distributed algorithms is considered inn-player games. We shall first formulate a new geometric problem for finding Pareto solutions. It involves solving joint tangents for the players' objective functions. This problem can then be solved with distributed iterative methods, and two such methods are presented. The principal results are related to the analysis of the geometric problem. We give conditions under which its solutions are Pareto optimal, characterize the solutions, and prove an existence theorem. There are two important reasons for the interest in distributed algorithms. First, they can carry computational advantages over centralized schemes. Second, they can be used in situations where the players do not know each others' objective functions.     

Minimizing the total completion time on a single machine with the learning effect and multiple availability constraints

This paper considers a single machine scheduling problem with the learning effect and multiple availability constraints that minimizes the total completion time. To solve this problem, a new binary integer programming model is presented, and a branch-and-bound algorithm is also developed for solving the given problem optimally. Since the problem is strongly NP-hard, to find the near-optimal solution for large-sized problems within a reasonable time, two meta-heuristics; namely, genetic algorithm and simulated annealing are developed. Finally, the computational results are provided to compare the result of the binary integer programming, branch-and-bound algorithm, genetic algorithm and simulated annealing. Then, the efficiency of the proposed algorithms is discussed.     

The Pareto fitness genetic algorithm: Test function study

Evolutionary algorithms have shown some success in solving multiobjective optimization problems. The methods of fitness assignment are mainly based on the information about the dominance relation between individuals. We propose a Pareto fitness genetic algorithm (PFGA) in which we introduce a modified ranking procedure and a promising way of sharing; a new fitness function based on the rank of the individual and its density value is designed. This is considered as our main contribution. The performance of our algorithm is evaluated on six multiobjective benchmarks with different Pareto front features. Computational results (quality of the approximation of the Pareto optimal set and the number of fitness function evaluations) proving its efficiency are reported.     

Algorithms for finding logical regularities in pattern recognition

Three algorithms for finding logical regularities of classes in the precedent-based recognition problem are proposed. Logical regularities of classes are defined as conjunctions of special oneplace predicates that determine the membership of a value of a feature in a certain interval of the real axis. The conjunctions are true on as large subsets of reference objects of a certain class as possible. The problem of finding logical regularities is formulated as a special integer programming problem. Relaxation, genetic, and combinatorial algorithms are proposed for solving this problem. Comparison results for these algorithms using model and real-time problems are presented. Comparison results for various estimate evaluation recognition algorithms that use logical regularities of classes in voting procedures are also presented.     

Polynomial time algorithms for two special classes of the proportionate multiprocessor open shop

This paper describes a complex scheduling problem taken from a hospital diagnostic testing center that schedules hundreds of patients in an open shop environment consisting of multiple facilities and multiple processors. This scheduling problem, known as the multiprocessor open shop (MPOS) problem, is strongly NP-hard with few published results. Realizing that in many MPOS environments processing times are stage-dependent, not both job and stage-dependent, this paper examines a new class of problems for the MPOS—proportionate ones. This paper exploits the structural nature of the proportionate MPOS and defines new terms. Despite the enormous complexity of the MPOS problem, this work demonstrates that polynomial time algorithms exist for two special cases. Since other applications of this problem exist in service and manufacturing environments, solving the proportionate MPOS problem is not only significant in the theory of optimization, but also in many real-world applications.     

Solving the knapsack problem with imprecise weight coefficients using genetic algorithms

This paper investigates solving the knapsack problem with imprecise weight coefficients using genetic algorithms. This work is based on the assumption that each weight coefficient is imprecise due to decimal truncation or coefficient rough estimation by the decision-maker. To deal with this kind of imprecise data, fuzzy sets provide a powerful tool to model and solve this problem. We investigate the possibility of using genetic algorithms in solving the fuzzy knapsack problem without defining membership functions for each imprecise weight coefficient. The proposed approach simulates a fuzzy number by distributing it into some partition points. We use genetic algorithms to evolve the values in each partition point so that the final values represent the membership grade of a fuzzy number. The empirical results show that the proposed approach can obtain very good solutions within the given bound of each imprecise weight coefficient than the fuzzy knapsack approach. The fuzzy genetic algorithm concept approach is different, but gives better results than the traditional fuzzy approach.     

Global Minimization Algorithms for Holder Functions

This paper deals with the one-dimensional global optimization problem where the objective function satisfies a Hölder condition over a closed interval. A direct extension of the popular Piyavskii method proposed for Lipschitz functions to Hölder optimization requires an a priori estimate of the Hölder constant and solution to an equation of degree N at each iteration. In this paper a new scheme is introduced. Three algorithms are proposed for solving one-dimensional Hölder global optimization problems. All of them work without solving equations of degree N. The case (very often arising in applications) when a Hölder constant is not given a priori is considered. It is shown that local information about the objective function used inside the global procedure can accelerate the search signicantly. Numerical experiments show quite promising performance of the new algorithms.     

Discrete space location-allocation solutions from genetic algorithms

Genetic algorithms are adaptive sampling strategies based on information processing models from population genetics. Because they are able to sample a population broadly, they have the potential to out-perform existing heuristics for certain difficult classes of location problems. This paper describes reproductive plans in the context of adaptive optimization methods, and details the three genetic operators which are the core of the reproductive design. An algorithm is presented to illustrate applications to discrete-space location problems, particularly thep-median. The algorithm is unlikely to compete in terms of efficiency with existingp-median heuristics. However, it is highly general and can be fine-tuned to maximize computational efficiency for any specific problem class.     

Molecular conformation on the CM-5 by parallel two-level simulated annealing

In this paper, we propose a new kind of simulated annealing algorithm calledtwo-level simulated annealing for solving certain class of hard combinatorial optimization problems. This two-level simulated annealing algorithm is less likely to get stuck at a non-global minimizer than conventional simulated annealing algorithms. We also propose a parallel version of our two-level simulated annealing algorithm and discuss its efficiency. This new technique is then applied to the Molecular Conformation problem in 3 dimensional Euclidean space. Extensive computational results on Thinking Machines CM-5 are presented. With the full Lennard-Jones potential function, we were able to get satisfactory results for problems for cluster sizes as large as 100,000. A peak rate of over 0.8 giga flop per second in 64-bit operations was sustained on a partition with 512 processing elements. To the best of our knowledge, ground states of Lennard-Jones clusters of size as large as these have never been reported before.Also a researcher at the Army High Performance Computing Research Center, University of Minnesota, Minneapolis, MN 55415     

Extensive Testing of a Hybrid Genetic Algorithm for Solving Quadratic Assignment Problems

A robust search algorithm should ideally exhibit reasonable performance on a diverse and varied set of problems. In an earlier paper Lim et al. (Computational Optimization and Applications, vol. 15, no. 3, 2000), we outlined a class of hybrid genetic algorithms based on the k-gene exchange local search for solving the quadratic assignment problem (QAP). We follow up on our development of the algorithms by reporting in this paper the results of comprehensive testing of the hybrid genetic algorithms (GA) in solving QAP. Over a hundred instances of QAP benchmarks were tested using a standard set of parameters setting and the results are presented along with the results obtained using simple GA for comparisons. Results of our testing on all the benchmarks show that the hybrid GA can obtain good quality solutions of within 2.5% above the best-known solution for 98% of the instances of QAP benchmarks tested. The computation time is also reasonable. For all the instances tested, all except for one require computation time not exceeding one hour. The results will serve as a useful baseline for performance comparison against other algorithms using the QAP benchmarks as a basis for testing.     

Simulated annealing and genetic algorithms for minimizing mean flow time in an open shop

This paper considers the problem of scheduling n jobs on m machines in an open shop environment so that the sum of completion times or mean flow time becomes minimal. It continues recent work by Bräsel et al. [H. Bräsel, A. Herms, M. Mörig, T. Tautenhahn, T. Tusch, F. Werner, Heuristic constructive algorithms for open shop scheduling to minmize mean flow time, European J. Oper. Res., in press (doi.10.1016/j.ejor.2007.02.057)] on constructive algorithms. For this strongly NP-hard problem, we present two iterative algorithms, namely a simulated annealing and a genetic algorithm. For the simulated annealing algorithm, several neighborhoods are suggested and tested together with the control parameters of the algorithm. For the genetic algorithm, new genetic operators are suggested based on the representation of a solution by the rank matrix describing the job and machine orders. Extensive computational results are presented for problems with up to 50 jobs and 50 machines, respectively. The algorithms are compared relative to each other, and the quality of the results is also estimated partially by a lower bound for the corresponding preemptive open shop problem. For most of the problems, the genetic algorithm is superior when fixing the same number of 30 000 generated solutions for each algorithm. However, in contrast to makespan minimization problems, where the focus is on problems with an equal number of jobs and machines, it turns out that problems with a larger number of jobs than machines are the hardest problems.     

Using duality to solve generalized fractional programming problems

In this paper we explore the relations between the standard dual problem of a convex generalized fractional programming problem and the partial dual problem proposed by Barros et al. (1994). Taking into account the similarities between these dual problems and using basic duality results we propose a new algorithm to directly solve the standard dual of a convex generalized fractional programming problem, and hence the original primal problem, if strong duality holds. This new algorithm works in a similar way as the algorithm proposed in Barros et al. (1994) to solve the partial dual problem. Although the convergence rates of both algorithms are similar, the new algorithm requires slightly more restrictive assumptions to ensure a superlinear convergence rate. An important characteristic of the new algorithm is that it extends to the nonlinear case the Dinkelbach-type algorithm of Crouzeix et al. (1985) applied to the standard dual problem of a generalized linear fractional program. Moreover, the general duality results derived for the nonlinear case, yield an alternative way to derive the standard dual of a generalized linear fractional program. The numerical results, in case of quadratic-linear ratios and linear constraints, show that solving the standard dual via the new algorithm is in most cases more efficient than applying directly the Dinkelbach-type algorithm to the original generalized fractional programming problem. However, the numerical results also indicate that solving the alternative dual (Barros et al., 1994) is in general more efficient than solving the standard dual.This research was carried out at the Econometric Institute, Erasmus University Rotterdam, the Netherlands and was supported by the Tinbergen Institute Rotterdam