The Coors warehouse scheduling problem involves finding a permutation of customer orders that minimizes the average time that customers' orders spend at the loading docks while at the same time minimizing the running average inventory. Search based solutions require fast objective functions. Thus, a fast low-resolution simulation is used as an objective function. A slower high-resolution simulation is used to validate solutions. We compare the performance of a constructive scheduling algorithm to a genetic algorithm and local search approach. The constructive algorithm is based on a heuristic built specifically for this application. We also tested a hybrid of the genetic algorithm and local search approaches by initializing the search using the domain-specific heuristic. This hybrid genetic algorithm was able to find the best solutions when evaluated by the high-resolution simulation. Finally, we consider the effect of using the high-resolution simulation to filter a set ...

Crossover operators that preserve common components can also preserve representation level constraints. Consequently, these constraints can be used to beneficially reduce the search space. For example, in flow shop scheduling problems with order-based objectives (e.g. tardiness costs and earliness costs), search space reductions have been implemented with precedence constraints. Experiments show that these (heuristically added) constraints can significantly improve the performance of Precedence Preserving Crossover--an operator which preserves common (order-based) schemata. Conversely, the performance of Uniform OrderBased Crossover (the best traditional sequencing operator) improves less--it is based on combination. Overall, the results suggest that conditions exist where Precedence Preserving Crossover should be the best performing genetic sequencing operator. 1 INTRODUCTION Due to their lower development cost, it is appealing to use domain independent search techniques (e.g. geneti...

"Survival of the fittest" is often seen as the driving force behind adaptation and evolution. For sure, all evolutionary algorithms use fitness-based selection. However, it is not necessary to know where you are, to know where you are going. Similarly, it is not necessary to know the fitness of a solution, to find a better solution. The GENIE algorithm uses random parent selection and a non-elitist generational replacement scheme. Experiments on a non-trivial instance of the Traveling Salesman Problem show that heuristic operators in GENIE can converge to the optimal solution without evaluating fitness. 1 Introduction A genetic algorithm (GA) has three basic features: a population of solutions, fitness-based selection, and crossover [Hol75][Gol89]. In isolation, fitness-based selection over a population of solutions increases the proportion of building blocks (schemata) with above-average fitness. Traditionally, crossover is used to recombine these building blocks into new solutions. ...

When employing evolutionary algorithms it is often the case that the evaluation function is the most computationally expensive part of the algorithm. Our evaluation function calculates the no fit polygon (NFP) for two polygons and then calculates the smallest convex hull for these two polygons. This process is repeated for each polygon. As the manipulation of polygons is computational expensive, the algorithm shows a bottleneck at this stage. However, many of the evaluations are simply reevaluating solutions that have already been evaluated. In order to use this information we use a cache which stores previous evaluations. By increasing the size of the cache size the speed of the algorithm is significantly increased. In addition the concept of a polygon type allows much better use to be made of the cache. In some circumstances, it may not be beneficial to use a cached evaluation. A reevaluation parameter is introduced which forces a complete reevaluation of a solution. We show that thi...

Modern high performance computing tasks cover a wide range of characteristics. Different programs offer greater or lesser opportunities to take advantage of spatial and temporal locality, threading, vectorization, and a range of computational demands. To cope with this wide range, future computers may require several heterogeneous computing elements, each with different capabilities and characteristics. In addition, as the nature of high performance computing changes, a real-time component may be added to the processing. The partitioning and scheduling of a such a task between these different elements is a complicated prospect, and we propose to apply genetic algorithms to aid in its solution. 1

vements we have shown that a `clashrich ' approach, working to solve constraint violations within a fixed-size timetable, outperforms `clash-spare' and `clash-free' methods that stretch a timetable in order to avoid violations; we have shown that a variety of local-hillclimbing mutation methods are needed and can work well [4]; we have shown that a variety of non-greedy but non-random initialisation methods can improve quality [10]; and we have shown that such techniques can scale successfully to tackle real-world problems, such as timetabling all of Edinburgh University's undergraduate exams or Kingston University's undergraduate lectures. Most recently [11] we have begun to explore an unusual phase-transition in generated problems involving only binary constraints; certain problems involving only a lowish density of constraints may be solved far less reliably than much more constrained problems. The problems in this phase-transition region seem to be very hard for non-evolutionary al

Abstract: Warehouses scheduling is the problem of sequencing requests of products to fulfill several customers ’ orders so as to minimize the average time and shipping costs. In this paper, a solution to the problem of multiple warehouses scheduling using the steady state genetic algorithm is presented. A mathematical model that organizes the relationships between customers and warehouses is also presented in this paper. Two scenarios of storage capacities (constants and varying capacities) and two strategies of search points (ideal point and random points) are compared. An analysis of the results indicates that multiple warehouses scheduling using the GENITOR approach with different warehouses capacities have better outcome than the usage of the traditional genetic algorithms).