Simulation optimization is a developing research area whereby a set of input conditions is sought that produce a desirable output (or outputs) to a simulation model. Although many approaches to simulation optimization have been developed, the research area is by no means mature. This research makes three contributions in the area of simulation optimization. The first is fundamental in that it examines simulation outputs, called "response surfaces," and notes their behavior. In particular both point and region estimates are studied for different response surfaces: Conclusions are developed that indicate when and where simulation-optimization techniques such as Response Surface Methodology should be applied. The second contribution provides assistance in selecting a region to begin a simulation-optimization search. The new method is based upon the artificial intelligence based approach best-first search. Two examples of the method are given. The final contribution of this research expands upon the ideas by Crouch for building a "Learner" to improve heuristics in simulation over time. The particular case of parameter-modification learning is developed and illustrated by example. The dissertation concludes with limitations and suggestions for future work. / Ph. D.

El objetivo general de esta tesis se centra en la investigación y desarrollo de algoritmos paralelos de búsqueda en grafos best-first search para arquitecturas multicore y cluster de multicore, que mejoran los existentes y se utilizan para resolver problemas de optimización combinatoria y de planificación, acompañado de un análisis de rendimiento (speedup, eficiencia, escalabilidad) de los mismos. La temática propuesta es de interés en la actualidad por la complejidad computacional de dichos algoritmos de búsqueda y las posibilidades que brindan las arquitecturas mencionadas. Los algoritmos presentados en esta tesis pueden aplicarse para resolver problemas reales como planificación de rutas óptimas, navegación automática de un robot o vehículo, alineamiento óptimo de secuencias, entre otros. Los temas de investigación derivados son múltiples y se refieren tanto a la paralelización de algoritmos sobre (a) arquitecturas de memoria compartida, como son los multicore (b) arquitecturas de memoria distribuida, como son los clusters (c) y también sobre arquitecturas híbridas, tal es el caso de los clusters de multicore. El aporte de la tesis es el desarrollo de dos algoritmos paralelos best-first-search propios, uno apto para su ejecución sobre máquinas de memoria compartida (multicore) y otro apto para máquinas de memoria distribuida (cluster), basados en el algoritmo HDA* (Hash Distributed A*), en los cuales se incluyen técnicas originales que optimizan su rendimiento. Asimismo, se presenta un análisis de rendimiento de los algoritmos desarrollados a medida que escala la carga de trabajo y la arquitectura paralela subyacente. Para finalizar, se compara la memoria consumida por ambos algoritmos y el rendimiento alcanzado cuando se los ejecuta sobre una máquina multicore; estos análisis presentan originalidad en el área. Los resultados arrojados indican que se obtendría un beneficio al convertir HDA* en una aplicación híbrida, cuando la arquitectura subyacente es un cluster de multicore, por lo que se sientan las bases para éste algoritmo híbrido.

With the technological advancement in computer science, Geographic Information Science (GIScience), and transportation, more and more complex path finding queries including category based queries are proposed and studied across diverse disciplines. A category based query, such as Optimal Sequenced Routing (OSR) queries and Trip Planning Queries (TPQ), asks for a minimum-cost path that traverses a set of categories with or without a predefined order in a graph. Due to the extensive computing time required to process these complex queries in a large scale environment, efficient algorithms are highly desirable whenever processing time is a consideration. In Artificial Intelligence (AI), a best first search is an informed heuristic path finding algorithm that uses domain knowledge as heuristics to expedite the search process. Traditional best first searches are single-variate in terms of the number of variables to describe a state, and thus not appropriate to process these queries in a graph. In this dissertation, 1) two new types of category based queries, Category Sequence Traversal Query (CSTQ) and Optimal Sequence Traversal Query (OSTQ), are proposed; 2) the existing single-variate best first searches are extended to multivariate best first searches in terms of the state specified, and a class of new concepts--state graph, sub state graph, sub state graph space, local heuristic, local admissibility, local consistency, global heuristic, global admissibility, and global consistency--is introduced into best first searches; 3) two bivariate best first search algorithms, C* and O*, are developed to process CSTQ and OSTQ in a graph, respectively; 4) for each of C* and O*, theorems on optimality and optimal efficiency in a sub state graph space are developed and identified; 5) a family of algorithms including C*-P, C-Dijkstra, O*-MST, O*-SCDMST, O*- Dijkstra, and O*-Greedy is identified, and case studies are performed on path finding in transportation networks, and/or fully connected graphs, either directed or undirected; and 6) O*- SCDMST is adopted to efficiently retrieve optimal solutions for OSTQ using network distance metric in a large transportation network. / Ph. D.