ABSTRACT. We present a method to solve chance-constrained problems with binary variables and a finite number of scenarios. Our approach starts with an infeasible system consisting of all scenario realizations, and by identifying irreducibly infeasible subsystems (IIS), we are able to remove constraints and find good candidate solutions for the original problem. The method consists in a branch-and-bound algorithm, and our strategy is to quickly find feasible nodes based on IIS for the relaxed formulation of the problem. We apply our technique to large scale instances of binary chance-constrained problems encountered in the literature.

ABSTRACT. Traditional stochastic programs assume that the probability distribution of uncertainty is known. However, in practice the probability distribution oftentimes is not known or cannot be accurately approximated. One way to address such distributional ambiguity is to work with distributionally robust convex stochastic programs (DRSP), which minimize the worst-case expected cost with respect to a set of probability distributions. In this paper we analyze the case where there is a finite number of possible scenarios and study the question of how to identify the critical scenarios resulting from solving a DRSP. We illustrate that not all, but only some scenarios might have ``effect" on the optimal value/solution. We propose easy-to-check conditions to identify effective and ineffective scenarios for that class of problems. Computational results show that identifying effective scenarios provides useful insight on the underlying uncertainties of the problem.

ABSTRACT. In Stochastic Programing a key issue is the generation of scenarios from a random variable. This is directly related to the quality of the solution of the problem but, as the number of scenarios increases, the complexity of the problem also increases. This is the principal motivation of scenario reduction techniques, which attempt to reduce the number of scenarios without worsening too much the quality of the solution. Most of the work in this area focuses on problems where the goal is to minimize the expected value of a function. In this talk we discuss some techniques that aim at generating scenarios when the goal of the problem is minimize a risk function that represents a decision maker’s aversion to risk. The techniques are illustrated with some numerical examples to demonstrate the impact of the methods.

ABSTRACT. We introduce a class of stochastic programming problems where scenarios are affinely parametrized by the vertices of a given polytope Q. Since the number of vertices is exponential with respect to the size of the description of Q, a natural integer programming formulation that includes binary variables to choose which scenarios are satisfied is simply impractical. Thus we present a formulation that discards the k worst scenarios for a given vector x without including variables for each scenario, leading to a mixed-binary program of moderate size. We also study a model where the average cost of the k worst scenarios is considered, for which we present a compact linear formulation. These models can be interpreted in the context of VaR- and CVaR-constrained problems, respectively, exhibiting similar relationships. A key tool we exploit is a dynamic program for linear optimization over a set of integral matrices with lexicographically ordered rows, which might be of independent interest.

ABSTRACT. This paper proposes the usage of different extensions of the well-known Periodic Vehicle Routing Problem (PVRP) for addressing an insular rural waste collection system design problem. All the proposed models minimize maritime transportation costs for a single barge to visit a set of islands in a 6-day period. This problem consists of selecting the docks of each island as collection sites, visit frequency and days for each selected dock, and daily visit sequence, where weight/volume capacity and time duration constraints are taken into account. Visit days and frequency decisions are modeled using a visit pattern approach, where a specific pattern must be assigned to each selected dock. The models are solved with Branch & Bound using a commercial solver for a set of real instances of different islands belonging to Chiloe and Palena provinces, Chile. Computational results are presented along with a sensitivity analysis considering some key parameters of the problem.

ABSTRACT. Este trabajo se enfoca en el problema de recolección de residuos domiciliarios generados en un conjunto de islas rurales ubicadas en el sur de Chile. En este estudio se buscan formulaciones para el problema, basadas en Programación Entera Mixta, los cuales integran decisiones de selección de puntos de recolección y visitas en cada isla (muelles), asignación de frecuencias y días de visita para cada muelle seleccionado, y decisiones diarias de ruteo vehicular (secuencia de visitas). Una particularidad de este problema es que se hace necesario identificar en qué casos es conveniente visitar una isla más de una vez, es decir utilizando más de un punto de recolección. De este modo, la presente propuesta se enfoca en realizar dicho análisis, sobre la base de enfoque Bi-Objetivo, el cual tendrá en cuenta tanto los costos de transporte marítimos, como también los costos terrestres al interior de las islas, correspondiente al traslado de basura hacia los puntos de recolección o muelles.

ABSTRACT. En este trabajo, se presenta una metodología para resolver el problema de programación de las rutas diarias de los camiones para una empresa farmacéutica. La metodología propuesta consta de tres etapas: (1) clusterización de los locales (2) resolver el problema de vehículo de reparto con capacidad limitada y ventanas de tiempo para cada uno de los clusters con el fin de determinar el número mínimo de camiones requeridos para satisfacer la demanda de cada cluster, (3) resolver el problema del vendedor viajero con ventanas de tiempo para cada uno de los recorridos con el fin de disminuir el tiempo de transporte de cada ruta entregada en la etapa anterior. Aplicando la metodología propuesta se logró una reducción de un 27% en los costos totales de transporte para la empresa en estudio.

ABSTRACT. El problema de Ruteo de Vehículos Eléctricos Multi Deposito (eMDVRP) es una variante del clásico MDVRP, agregando restricciones relacionadas con el uso de los vehículos eléctricos, los cuales pueden recorrer una distancia limitada desde el deposito, pero el vehículo puede ir a un punto de recarga donde puede ser recargado (total o parcialmente) para incrementar la distancia máxima a recorrer. Se presentan vecindarios eficientes, que luego son usados en metaheurísticas de trayectoria. La generación de nuevos vecindarios es necesaria debido a que el uso de operadores clásicos no resulta adecuado por la cantidad de vecinos infectable que genera. Se hace un estudio comparativo de los vecindarios y de las metaheurísticas. Adicionalmente, para verificar el correcto funcionamiento, de forma eficiente y eficaz, de los algoritmos implementados se considera el caso cuando el vehículo inicia con infinita energía desde el deposito, transformando el problema en el MDVRP clásico.

ABSTRACT. The Vehicle Routing Problem (VRP) is one of the main combinatorial problems that has several applications in Industrial Engineer, such as: distributions of goods, routing emergency vehicle, etc. In this work we studied the computational performance of the most used MS Excel Solvers (IBM ILOG Cplex, Opensolver, Lindo What’s Best) applied to solve different versions of the VRP: Homogenous and Heterogeneous Capacitated VRP, Homogeneous and Heterogeneous Capacitated VRP with Time Windows. The performance was measured in different scale problems by the quality of the solution obtained and the required computing time.

ABSTRACT. A Mixed-Integer Linear Programming model is proposed to determine the optimal number, location and capacity of the warehouses required to support a 5-year forecast for a business with seasonal demand. Discrete transportation costs, dynamic warehouse contracting, and handling of safety stock are the three main features that make the problem different from the previous reported work in the facility location problem. The most effective modeling alternative for each of these complicating issues was identified. To solve large-scale problems, tightening constraints, approximating reformulations and multistage heuristics are proposed. A 2-stage heuristic based on products grouping is able to obtain the optimal solution in 95% of the instances tested. A 3-stage heuristic, using a single mode model as intermediate step, was found to be better suited for very large-scale problems yielding only a 1% deviation from the optimal solution.

ABSTRACT. Este trabajo estudia problema dinámico de localización y despacho de vehículos de emergencias médicas. El objetivo es maximizar la cantidad de servicios atendidos en un tiempo menor o igual al establecido, para múltiples tipos de emergencias. La matheurística propuesta busca ser la base de una herramienta cuantitativa, que pueda ser implementada en la práctica. El problema de localización es formulado con un enfoque vértices y resuelto de manera exacta. El problema de despacho, es resuelto a través de una heurística basada en del índice de preparación. Finalmente, el problema relocalización parcial o total de flota es resuelto de manera exacta con un modelo de dos etapas. En la primera, se soluciona un modelo matemático orientado a la maximización de cobertura, en la segunda, se minimiza el máximo tiempo de desplazamiento de los vehículos a relocalizar. Los experimentos preliminares muestran resultados prometedores con instancias basadas en información real.

ABSTRACT. New alternative energy sources, such as biomass, have been gaining importance and relevance along the time. Biofuels are obtained from biomass, the energy from biological material, through complex processes that take place in a multifunctional facility: the biorrefinery. Our paper proposes a procedure to properly locate a biorrefinery in Navarra (Spain) from economic and environmental point of view. Thus, a Mixed Integer Linear Programming (MILP) model coded in GAMS is developed to solve a multiobjective Facility Location Problem (FLP). Epsilon-Constraint methodology is followed to build up the Pareto Frontier (PF) along with the aforementioned criteria. Promising results are obtained allowing us to determine the best area to pinpoint a biorrefinery as well as to define Pareto optimal decisions.

ABSTRACT. El presente trabajo analiza la eficiencia de las sucursales de una entidad financiera aplicando la metodología Data Envelopment Analysis, y su evolución a lo largo de 4 semestres consecutivos mediante el índice de Malmquist. Con este objetivo y trabajando con 17 indicadores para 51 sucursales de la organización, se aplicó inicialmente un modelo CCR input orientado cuyos resultados no permitieron distinguir satisfactoriamente entre unidades eficientes e ineficientes. A fin de mejorar la calidad de los resultados se aplicó el método IO-Stepwise de Lins y Moreira (1999) para reducir dimensiones, seleccionándose un total de 9 indicadores (3 entradas y 6 salidas) que representan adecuadamente el problema a abordar. Con estas variables se logró analizar la evolución de la eficiencia para los subperíodos considerados, distinguiendo entre el cambio en el uso de los recursos y el desplazamiento de la frontera eficiente, y efectuar recomendaciones a la empresa en base a los resultados obtenidos.

ABSTRACT. El pronóstico de la demanda del consumo eléctrico mediante series de tiempo de alta frecuencia es un área de aplicación muy relevante para investigadores en modelos de pronóstico, debido a su importancia en la gestión operacional de los sistemas energéticos. Sin embargo, modelos de aprendizaje de máquinas, tales como las Redes Neuronales o Support Vector Regression (SVR), no han tenido un gran desarrollo para este tipo de aplicaciones, debido a la dificultad que resulta especificar este tipo de modelos de forma efectiva y parsimoniosa. En este trabajo se extiende el método Kernel-Penalized SVR (KP-SVR) al análisis de series de tiempo con el fin de determinar de forma automática el número de rezagos adecuados en el proceso autorregresivo. Este método realiza una eliminación de atributos hacia atrás al ajustar de forma iterativa el ancho de una función de kernel no isotrópico. El modelo desarrollado es aplicado a la industria eléctrica.

ABSTRACT. In this talk we address the problem of representing in a given bounded region a set of individuals, which has attached a dissimilarity measure and a statistical value interpreted as the magnitude of the individual. This problem, which extends the standard Multidimensional Scaling Analysis, is written as a global optimization problem whose objective is the difference of two convex functions (DC). Suitable DC decompositions allow us to use the DCA algorithm in a very efficient way. Our algorithmic approach is used to visualize two real-world datasets, both static and dynamic.

ABSTRACT. En este trabajo se presenta el diseño y aplicación de una metodología basada en la clusterización de series de tiempo utilizando modelos de variables latentes para descomponer y describir el mercado de clientes regulados de una compañía de distribución eléctrica chilena durante un período de estudio que abarca los años 2013 y 2014, para ello se propone utilizar la herramienta estadística de Análisis de Componentes Independientes, en particular se propone utilizar el algoritmo fastICA para encontrar distintas combinaciones de componentes independientes dependiendo de cada experimento y luego seleccionar aquel resultado que mejor permita explicar las señales de consumo observadas e interpretar dichos componentes como patrones de consumos asociados a grupos de clientes individuales, dicha herramienta fue aplicada a dos data sets provenientes de equipos de medición asociados a transformadores y a subestaciones.

ABSTRACT. Given the increasing dominance of natural gas in the U.S. electric power systems, the traditional concern of electricity regulators to ensure that market agents take efficient power plant investment decisions expands to pipelines in the gas system. Ensuring pipeline capacity, particularly when under tight supply conditions, involves entering into very long-term firm transportation contracts, and therefore introduces a major source of risk for generators. In this paper we analyze the problem of gas and electricity long-term planning coordination and the security of supply consequences, applied to New England. We assess how a risk-averse gas-fired generator owner underinvests in pipeline capacity when no hedging tools are available, and to what extent the gap could be bridged by adding long-term financial markets for risk. For this purpose, we formulate a market-equilibrium under four different settings solved by means of a Mixed Complementarity Problem.

ABSTRACT. Do national leaders influence stock markets? We use random exits of leaders while in office to determine whether these exogenous transitions are associated to changes in stock market returns across countries. We focus on the cases in which the leader's rule ended due to either a sudden accident or a long-standing illness. We find that an accident does not trigger a significant price reaction or change in volatility. A long-standing illness, however, is associated to a positive and significant change in cumulative average abnormal returns, but not to a change in volatility. Using data on rumors about the leaders' health, we find that the start of a string of rumors is associated to a significant drop in average abnormal returns, but no change in volatility. This suggests that leader's ailment generates expectations of deficient rule. Further analysis suggests that the significant drop-and-rebound effect is strongest in autocratic regimes.

ABSTRACT. We analyze the role of government leaders and the effects of political uncertainty on innovation, as measured by growth in patent applications. Utilizing real options reasoning, we argue that patents confer firms' two options: they allow delaying investments required to exploit new technologies and they also confer an option to grow in the future. We posit that firms react to the political uncertainty caused by the unexpected death of a head of state by increasing patenting, and that these increases are contingent on the personal characteristics of the departed. We use an expanded set of random leadership transitions—caused by natural death or terminal illness—to show that heads of states are responsible, on average, for an increase of approximately 10 percentage points in the growth of patent applications. We also show the departure of leaders with low levels of professional attainment and with judicial records further increased patenting.

ABSTRACT. A life-cycle problem from the perspective of the individual is considered. In this problem an individual receives a periodic income and decides how much to consume and invest for the future. The objective is to maximize periodic consumption’s benefits which are modeled by convex risk-measuring functions, such as utility functions. The uncertainty comes from the income and the return on investment. In this talk we formulate a life-cycle problem as a multi-stage stochastic program, which allows us to model the problem’s dynamics without the strong assumptions made in literature. Since the number of scenarios is very large, we solve the problems using the Stochastic Dual Dynamic Programming algorithm. We present numerical results and use existing Economics models as benchmark to compare optimal policies. Finally, a more complex model that cannot be solved by standard solution methods is presented, showing the output of the multi-stage model as policies that may be implemented.

ABSTRACT. Os setores industriais possuem uma dinâmica de desenvolvimento que pode depender tanto de suas características intrínsecas como o cenário econômico que os afeta tanto quanto as Instituições de Ciência e Tecnologia voltadas ao seu desenvolvimento e competitividade. Este trabalho propõe uma multimetodologia que apoie a decisão de priorizar os setores industriais para balizar as estratégias de investimento em ciência e tecnologia considerando dados qualitativos e quantitativos através de series históricas e pesquisa de campo. Para a estruturação do problema foi utilizada a abordagem Value Focused Thinking (VFT) e para a priorização foi utilizado o método Simple Multi-Attribute Rating Technique (SMART). A utilização desses métodos e o estudo dos cenários permitiram a priorização dos setores industriais a fim de se balizar os investimentos da ICT considerando algumas das possíveis variações econômicas.

ABSTRACT. En este trabajo intentamos evaluar el bienestar social de 148 países, contemplando un conjunto de indicadores que buscan captar el estándar de vida de los ciudadanos, su acceso a la salud y a la tecnología, sus logros educativos, el impacto en el medio ambiente y la calidad de la gobernabilidad. En una primera etapa, utilizamos un método de clasificación robusto (K-medias recortado), obteniendo tres grupos de países con características similares respecto de los indicadores seleccionados. En una segunda etapa, recurrimos al método multiatributo TOPSIS para ordenar los países de cada grupo en sentido decreciente de su grado de bienestar social y de esta manera obtener un ranking de todos ellos. Finalmente, comparamos los resultados obtenidos con el Índice de Desarrollo Humano publicado por el Programa de las Naciones Unidas para el Desarrollo.

ABSTRACT. A região semiárida brasileira enfrenta o seu maior período de estiagem dos últimos 100 anos, com chuvas abaixo da média histórica. A situação não é diferente no Estado do Rio Grande do Norte (RN), que possui grande parte de seu território situado em área de semiárido. Diante de tal contexto, a gestão de recursos hídricos e o combate aos efeitos da seca ganham ainda mais importância quanto à minimização dos impactos sofridos pela população. As decisões tomadas nesse âmbito podem considerar múltiplos critérios avaliados por mais de um tomador de decisão. Este estudo tem como objetivo propor um modelo multicritério para tomada de decisão em grupo quanto à priorização de municípios da atingidos pela estiagem. O modelo foi aplicado para o caso dos municípios situados no RN e, a partir de tal aplicação, os gestores públicos podem direcionar seus esforços quanto às ações de combate à seca.

ABSTRACT. This paper puts forward a framework to support the process of allocating projects in line with an organizations restrictions by using mathematical programming. The model was formulated after reviewing the literature and being guided by the findings of searches; an MCDA method was used to classify projects and PMs (first stage) and mathematical programming to allocate projects (second stage) and subsequently information on the learning effect was added as a variable to the model. An application of the proposed model was implemented at a Brazilian electric energy company. The results demonstrated that projects and project managers can be classified into definable categories, thereby enabling the process of project allocation to be undertaken more effectively and doing so in a systematic and more efficient way. The proposed model can support an organization by allocating its most critical projects to its best qualified and experienced professionals whose performances have been successful.

ABSTRACT. La dificultad que se presenta en el problema de programación flexible de tareas, es ampliamente conocida en el área de la optimización combinatoria y en la familia de problemas de scheduling. Diseñar un algoritmo computacionalmente eficiente, que permita resolver diversos conjuntos de instancias para este problema, continúa presentando un desafío. Diversas técnicas han sido empleadas, sin embargo, generar algoritmos especializados sobre subconjuntos de instancias presenta un enfoque menos abordado. Mediante computación evolutiva y en particular mediante programación genética, se generan algoritmos capaces de minimizar el tiempo de procesamiento del conjunto de trabajos sobre máquinas factibles. Los resultados muestran que la generación automática de algoritmos resulta una técnica promisoria al ser aplicada sobre subconjuntos de instancias, encontrando una variedad de algoritmos especializados.

ABSTRACT. Neste trabalho são apresentadas duas versões da metaheurística ILS para o problema de programação de máquinas paralelas com minimização do atraso ponderado. Para cada tarefa é conhecido o tempo de processo, a data de entrega e o peso por dias de atraso. O objetivo é alocar as tarefas às máquinas minimizando a soma dos atrasos ponderados. Para otimizar cada máquina, foram usadas buscas locais do tipo swap - a troca da posição de duas tarefas, e twist - a inversão da ordem de um conjunto de tarefas. Na versão ILS-Clássica, a otimização entre máquinas se dá por movimentos de realocação e troca de tarefas. Na segunda versão, foi utilizada a técnica Very Large-scale Neighborhood Search-VLNS. Foram realizados testes computacionais com problemas benchmak, cujas soluções ótimas são conhecidas e com problemas maiores, para os quais não são conhecidas as soluções ótimas. Assim, foi possível verificar que a versão ILS-VLNS foi mais eficiente na resolução do problema do que a versão ILS-Clássica.

ABSTRACT. El estudio de los problemas de secuenciamiento de actividades es aún un campo vigente en la industria y la academia. Actualmente, los problemas que incorporan el uso de energía han cobrado prioridad. Así, este trabajo está enfocado en la evaluación de métodos meta-heurísticos que buscan soluciones energéticamente eficientes a los problemas de programación de actividades multimodales con recursos restringidos (MRCPSP). Este estudio tiene tres objetivos, primero se propone el MRCPSP-ENERGY, que consiste en un problema con un solo objetivo que tiene en cuenta el consumo de energía y el makespan de manera simultánea. Segundo, se propone una nueva librería de prueba PSPLIB-ENERGY (disponible en http://gps.webs.upv.es/psplib-energy/) que consiste en una extensión de la conocida librería PSPLIB. Y tercero, se propone un algoritmo genético basado en las mejores metodologías para resolver el MRCPSP, con el fin de brindar una primera aproximación de solución a la nueva librería propuesta.

ABSTRACT. O trabalho apresenta uma nova variante do Caixeiro Viajante denominada de Caixeiro Viajante Alugador com Passageiros (PCV-AlPa) e descreve um algoritmo Memético e um algoritmo GRASP (Greedy Randomized Adaptive Search Procedure) para sua solução. Os algoritmos desenvolvidos são validados através de um experimento computacional e submetidos à uma análise estatística. Ao final são tecidas conclusões sobre os resultados alcançados tanto em relação ao problema abordado e seu grau de complexidade, quanto em relação ao desempenho dos algoritmos propostos.

ABSTRACT. En el presente trabajo se desarrolló un modelo de simulación de eventos discretos que permitió evaluar el impacto que tiene sobre algunas variables de desempeño diferentes políticas de relocalización en un Servicio de Emergencias Médicas de Medellín, Colombia. Se realizó un análisis estadístico de la información de la operación del SEM de los años 2012 a 2014. Dicha base de datos contiene los registros del proceso de atención de 261818 solicitudes de servicios. Los resultados indican que al ubicar los vehículos en los barrios donde se prestó el último servicio, se presentan menores porcentajes de cancelaciones, tiempos de espera y permite una mayor disponibilidad de los vehículos para atender las solicitudes de los pacientes, debido a que están disponibles inmediatamente después de terminar un servicio. Lo anterior, además de tener un impacto positivo en los costos operacionales y la calidad del servicio, incrementa la probabilidad de supervivencia de los pacientes.

ABSTRACT. Los servicios de salud en cualquier parte del mundo deben asegurar una atención oportuna, segura y de alta calidad (Quality of Care). Por lo tanto, las entidades promotoras de salud en Colombia (EPS) deben garantizar procesos fáciles y ágiles en los servicios ofrecidos, evitando riesgos que pueden afectar la vida de los pacientes. Junto con lo anterior, en Colombia se definió la ley anti-tramites, siendo el cumplimiento de esta ley una problemática para las EPS ya que prestan servicios a más de 250.000 usuarios en ciudades como Medellín (Colombia). Esta aglomeración de pacientes ocasiona quejas e inconformidades de parte de los afiliados por las largas filas y demoras en los trámites que se presentan en el punto de atención al usuario. Para tal fin, la entidad desea disminuir los tiempos de atención, para evitar una sanción por parte de la superintendencia de salud colombiana, empleando como herramienta la simulación y realizar una propuesta eficiente que disminuya los costos y posibilidades de sanción.

ABSTRACT. The main objective of the paper is to show the power of Discrete Event Simulation in the design of clinical facilities, which will enable to build or adapt these centers achieving the expected Key Performance Indicators, such as standard waiting times according to acuity, average stay times and others. An example is presented where in order to minimize the average time of patients at a projected Emergency Unit of a public hospital in Chile, the operation of the adult care area is analyzed by using a simulation model. To do this, the areas involved in the process are identified: admission, triage, medical evaluation, procedures and treatments, which are modeled based on the seven steps methodology. The computational model is built and validated using expert judgment and supporting statistical data. Then, three scenarios are studied, resulting one of them in a 50% decrease in the average cycle time of patients compared to the original model, by modifying the patient´s attention model.

ABSTRACT. Se presenta una herramienta de fácil uso para generar y articular una malla curricular de una carrera universitaria y un programa de postgrado. Ésta se basa en un modelo matemático que describe formalmente el problema, integrando entre otras restricciones, las precedencias y los créditos transferibles (SCT-Chile) de cada asignatura y por cada semestre, según la definición adoptada por el Consejo de Rectores de las Universidades Chilenas (CRUCH) y la Comisión Nacional de Acreditación (CNA). El modelo es programado en GAMS 2.50 utilizando CPLEX 12.6.3, mientras que los parámetros de entrada y salida son entregados a través planillas de cálculo. Para ilustrar su uso, se presenta el caso de Ingeniería Civil Industrial y su articulación con el Magíster en Ciencias de la Ingeniería, Mención Ingeniería Industrial de la Universidad de Santiago de Chile. Se analizan diferentes escenarios obteniendo como resultado una nueva malla curricular que cumple con todos los requisitos.

ABSTRACT. In private higher education institutions, the management of teachers is critical to ensuring the quality of courses and to obtain good levels of performance across the education ministry. With the shortage of teachers with good degrees and low availability to occupy positions with partial or full time, the choice of teachers for each course can become very complex according to the size of the institution and the number of courses involved. In this paper we propose a mathematical model able to provide good quality solutions to this problem. The model was tested using data from a private institution of higher education showing its potential in bringing improvements to this process.

ABSTRACT. A gestão acadêmica tem se tornado um dos grandes temas discutidos nas Instituições de Ensino Superior - IES, em especial entre as privadas. Em um mercado concorrido, em que o número de IES cresce a cada ano, ser estratégico e se preocupar com uma efetiva gestão da qualidade e dos custos pode garantir não só a sobrevivência da instituição, mas também a sua renovação como negócio e sua sustentabilidade a longo prazo. Para isso, é importante que os processos para tomada de decisão utilizem técnicas e ferramentas capazes de oferecer soluções que gerem vantagens competitivas. Neste trabalho, foi desenvolvido um modelo matemático para alocação de recursos em um evento acadêmico de grande porte, com foco na otimização da utilização dos espaços disponíveis e consequente redução de custo. Foram utilizados os dados de uma IES privada e os resultados comprovam a eficácia da técnica, com redução de 63,5% das janelas de tempo com uma estrutura pelo menos 32% menor do que a utilizada sem o modelo.

ABSTRACT. The program ``Exactas va a la escuela'' (EVE) of the School of Sciences of the UBA offers to secondary schools a set of talks given by lecturers from the university. At the begging of the year, each school postulates its preferences for some of the available lectures and the goal of the program is to assign at least one lecture to each school at some time of the year. Each lecturer offers his/her activity during some limited days and times within the year and each school imposes similar time constraints. Also, geographical constraints are given by the lecturers restricting the schools they may visit. Finally, each lecturer may state a maximum frequency for his/her lecture to be given (once a week, twice a month, etc.). The assignment must satisfy all these constraints. In this work, we introduce this problem as the Lecturers Assignment Problem and we present an integer programming formulation to solve it showing the obtained results for real instances for the years 2015 and 2016.

ABSTRACT. El modelo de decisión desarrollado permite a estudiantes de Ingeniería Química entrenarse para la toma de decisiones en temas relacionados con la GPO dado que la simulación permite experimentar con una abstracción de un proceso sin afectarlo ni correr con los riesgos y costos que conlleva experimentar con el sistema real

ABSTRACT. As variáveis utilizadas para os problemas ambientais decorrem de fenômenos tangíveis, intangíveis e goodwill. A fim de estudar o comportamento desses critérios, utilizando ferramentas do MCDM, o Processo de Análise Hierárquica (AHP) apresenta-se como ferramenta prática para os tomadores de decisão que enfrentam tais problemas de priorização. Este trabalho é direcionado a um estudo de decisões relacionadas à Transposição do Rio São Francisco, onde as aplicações dos dados foram feitas no programa Expert Choice. Esse método mostrou ser eficiente para o processo de tomada de decisão, determinando a influência que os critérios tiveram sobre o resultado final. Portanto, a aplicação mais eficaz para esses problemas é aquela que separa os subcritérios, de acordo com o seu respectivo critério estratégico, para obter uma melhor visualização da tomada de decisão.

ABSTRACT. Design of personnel teams has been studied from diverse perspectives; the most common are the people and systems requirements perspectives. All these point of view are linked, which is the reason why it is necessary to study them simultaneously. Considering this gap, a decision making model is developed based on the main factors, models, and requirements mentioned in the literature. The findings indicate that the Personnel Behavior Based Lean model (PBBL) can be converted into a decision making model to select members. The study is focused not only on the individual candidate´s knowledge, skills, and aptitudes but also on how the model considers the company requirements, and the importance of each person to the project. A mathematical model is developed in order to find a team for a continuous improvement Lean project.

ABSTRACT. The variation of the Randic index of a graph is defined as sum of weights of all edges of graph. The weight of one edge is inverse value of the maximum of degrees of its end vertices. Let G(k,n) be the set of connected simple n-vertex graphs with minimum vertex degree k. In this paper we found in G(k,n) graphs for which the variation of the Randic index attains its minimum value. When k is less or equal to n/2 the extremal graphs are complete split graphs K(k,n-k), which have only vertices of two degrees, i.e. degree k and degree n-1, and the number of vertices of degree k is n-k, while the number of vertices of degree n-1 is k. For k greater or equal to n/2 the extremal graphs have also vertices of two degrees k and n-1, and the number of vertices of degree k is n/2. Further, we generalized results for graphs with given maximum degree.

ABSTRACT. O Fluxo de Potência Ótimo Reativo (FPOR) é um problema de otimização não linear de grande porte com variáveis contínuas e discretas, cujo objetivo é a minimização de um critério associado à rede de transmissão, sujeito às restrições físicas e operativas do sistema de potência. Em geral, o FPOR é resolvido por métodos que mesclam métodos de otimização discreta (tais como métodos de enumeração implícita e explícita) com métodos de otimização não linear. Dependendo da dimensão do problema, essa mescla de métodos pode se tornar computacionalmente inviável. Para resolver tal questão, alguns métodos foram propostos em que as variáveis discretas são tratadas de forma contínua, porém são acrescentadas à função objetivo, funções que penalizam valores não discretos para essas variáveis. Este trabalho propõe uma nova função de penalidade com objetivo de tratamento das variáveis discretas. O desempenho desta função é comparado ao de outras funções propostas na literatura para esta finalidade.

ABSTRACT. The Closest String Problem (CSP) that arises in computational molecular biology, coding theory and web searching is to find a string that minimizes the maximum Hamming distance from a given set of strings, the CSP is NP-hard problem. Several approximation and exact algorithms have been proposed for the problem to achieve optimal solutions using Mixed Integer Linear Programming. This paper proposes a new algorithm for the problem, based on dynamic programming. The algorithm is compared with an integer programming formulation for CSP. Furthermore, computational experiments in comparison tables will show the effectiveness of the proposed algorithm.

ABSTRACT. In this contribution we analyze the problem of finding a vertex coloring of a graph with the additional property that there are a fixed set of edges whose endpoint can have the same color. We say that these edges can be \emph{violated} with respect to the usual property of coloring in a graph. We present an integer programming formulation of this problem and begin with the study of the convex hull of integer solutions in it. These formulation involves the standard constraints for coloring in a graph and therefore we are forced to study this polytope for the families of graphs herein considered. We obtain the dimension, the minimum inequality system and some exponentially sized group of valid inequalities that we strongly believe suffice to describe the chromatic violation polytope for small instances of complete graphs with one violable edge.

ABSTRACT. Neste trabalho, estuda-se um problema de roteamento de veículos que integra as típicas decisões de programação e roteirização com a definição do tamanho da tripulação a ser designada em cada rota. Este problema surge na distribuição de produtos em centros urbanos congestionados onde os clientes são atendidos de forma agrupada e, portanto, os tempos de serviço nos grupos de clientes podem ser reduzidos ao incluir entregadores adicionais nas rotas. O problema é abordado por meio de uma matheurística de duas fases que integra duas abordagens metaheurísticas e uma formulação por partição de conjuntos do problema. Experimentos computacionais mostram a eficiência da matheurística ao ser capaz de gerar soluções de boa qualidade de forma relativamente rápida quando comparada com métodos da literatura.

ABSTRACT. When a traffic accident occurs in cities with a large traffic flow the roads surrounding the crash site are affected by traffic congestion. In some countries, such as Mexico, even small accidents are troublesome due to the fact that a claim adjuster from the car insurance company must arrive to the site and document the accident before the vehicle may be removed as required by law. Thus, in this particular setting, the location of the adjusters becomes a key factor in providing timely service. In this work, two integer programming models are introduced. With the aim of minimizing average response time, the first model considers allocations from each demand point to each adjuster. To simplify the problem, the second model considers allocations from each demand point to its k nearest adjusters. An efficient scatter search heuristic with path relinking that exploits model properties is proposed and assessed over a wide set of instances with very good results.

ABSTRACT. The metaheuristic method Continuous-GRASP solves efficiently general constrained global continuous optimization problems (Facó, Resende and Silva - 2011, 2012, 2013, 2014, 2015) by adapting the greedy randomized adaptive search procedure (GRASP) for discrete optimization to the case of constrained continuous variables. A new version that considers also discrete variables is presented - Generalized-GRASP - including integer variables. Small and medium scale MINLPs are currently addressed using continuous relaxations and solved by a branch-and-bound procedure. Large-scale instances cannot be solved this way due to the curse of dimensionality. Generalized-GRASP doesn't do any relaxation. GRASP random search and local improvement phases use independently a discrete and a continuous set. Linear and/or nonlinear constraints are incorporated in the objective function by quadratic penalty terms as in C-GRASP. Numerical solutions to difficult MINLP problems are presented.

ABSTRACT. In this work we address the one-to-one multi-commodity pickup and delivery traveling salesman problem (m-PDTSP), a challenging variant of the TSP which adds the need to transport commodities between locations. Each commodity has a weight, a pickup location, and a delivery destination. The goal is to find a minimum-cost tour such that all commodities are delivered to their destinations and the maximum capacity of the vehicle is not exceeded. The current literature on exact methods for the m-PDTSP typically focuses on mixed-integer programming, including Benders decomposition techniques and branch-and-cut. We propose a novel exact constraint programming formulation enhanced with discrete relaxations based on multivalued decision diagrams. Experimental results show that our approach outperforms the state-of-the-art methodologies for the m-PDTSP, closing several open instances from the literature.

ABSTRACT. La minería desempeña un importante rol en la economía chilena con un 11,2% del PIB nacional por lo que requiere una mejorar la eficiencia operacional. Se propone una mejora en el proceso gestión de cátados de cobre, que son almacenados en bodegas acorde a las necesidades, siendo retirados por camiones especializados. El estudio se centra en la carga de los paquetes a los camiones, donde se detectó falencias, con el incremento de costos, altos tiempos de estancia de los camiones en la empresa, el efecto de este problema se puede extender al nivel de servicio brindado tanto a clientes nacionales como internacionales. La propuesta de solución al problema se basa en el rediseño del proceso de recepción, carga y salida de los camiones en el centro, utilizando tecnologías RFID en los paquetes de cátados con la información para el proceso de carga y documental, permitiendo así mejor eficiencia operacional, un aumento en la capacidad de repuesta y un incremento en la rapidez de faena.

ABSTRACT. Staggered work hours are achieved by varying the starting and release times in firms. It consists of staggering worker’s starting times in intervals (e.g. 15 minutes), and adjusting their end times accordingly. In this paper, a bi-level formulation to model staggered work hours is presented. The behavior of the firm is modeled in the upper level as a minimization of the profit loss as a result of the firm’s productivity loss by means of using an agglomeration on time production function (production effects). The behavior of the workers is modeled in the lower level as the maximization of their utility function which depends on of their departure time and route choice decisions embedded in a dynamic user equilibrium model. The resulting model can be presented as a Mathematical Program with Equilibrium Constraint (MPEC). The MPEC model is used to provide insights about the interaction between workers and the firm, and to test scenarios in which incentives to stagger work arrival times.

ABSTRACT. During recent years, horizontal collaboration in logistics has gained attention due to the potential benefits that can be achieved such as cost reduction, increasing fulfillment rates and decreasing CO2 emissions due to reductions in travelled distances. However, successful real-world cases are rare since horizontal cooperation in logistics is usually not sustainable. One reason is that the companies often are competitors for sales and purchasing taking place after and before any collaboration. Two other are issues with trust and the lack of proper coordination mechanisms. This paper pays attention to this paradox of the lack of cases and discusses 33 identified practical issues that could explain this phenomenon. We propose four categories to classify the practical issues following a supply chain perspective. These are (i) design, (ii) planning and operations, (iii) market/business and (iv) behaviours. Furthermore, measures to mitigate these problems are proposed and discussed

ABSTRACT. No ano de 2011, a Fragata União, navio de guerra da Marinha do Brasil (MB), foi incorporada à Força Interina das Nações Unidas no Líbano (UNIFIL). Durante a missão, observou-se a necessidade de um modelo de previsão de demanda de peças sobressalentes para manutenção dos equipamentos de bordo mais adequado à nova realidade operacional do navio. Analisadas as demandas a partir de uma série histórica dos anos de 2012, 2013 e 2014, identificou-se que a demanda da maioria das peças apresentava um comportamento de acordo com a Curva Normal. Todavia, outras passavam longos períodos sem demanda, apresentando um comportamento severamente estocástico. O presente estudo utilizou os métodos de Croston e Syntetos-Boylan Approximation (SBA) comparando-os com o modelo utilizado atualmente pela MB. O estudo mostra-se relevante por ser totalmente inédito na MB e possuir um caráter dual, podendo ser adaptado em outros setores produtivos que necessitem de itens sobressalentes de elevado valor agregado.

ABSTRACT. Flower exports constitute an important sector in the economy of several Latin-American countries. Only in Colombia, flower exports were around $US1300 billion. Despite of this importance, export flowers still suffer from quality problems that not only depend on the planting and processing conditions but also on the logistics processes, represented in the so-called "cold chain" from the greenhouses to their final destination. The failure to maintain the cold chain is mostly due to waiting times. For such a reason Lean Manufacturing techniques have been encouraged by grower’s associations to streamline the process and reduce waiting times. The major drawback is that Lean techniques are a collection of, in many cases, unrelated practices and a full implementation can be overwhelming. This research explores, via Discrete Event Simulation and Experimental Design, how such techniques can be implemented. The conclusions have been endorsed by the Colombian grower’s association.

ABSTRACT. The objective of this study was to address the problem of planning the production and logistic operations of the agricultural and industrial activities in the tomato processing industry. We have developed a mathematical optimization model to support tactical decisions, which are the sizing of areas cultivated with tomatoes, the transportation from the growing regions to processing plants, the production lot sizing of concentrated tomato pulps and final products to consumers, as well as, the transport and inventories of these products in the supply chain. The mixed integer programming model presented here represents appropriately the main agricultural and industrial decisions and proved to be useful to generate optimized production and logistic plans for the tomato processing industry.

ABSTRACT. The Production Planning in a meat packing plant (MPP) involves determining the production levels of meat products and by-products in order to meet demand. The main operative action in a MPP is to cut up the carcasses (body of the animal) through the use of pre-established cutting patterns. Demand is given according not only to a meat product but also to a specific quality. Exist different grades of quality to carcasses, the demand of a specific product can be satisfy with the required quality or a better quality, in order to decrease unsatisfied demand and increase service level. A mixed integer linear programming model is used to assess the effect of demand substitution in the rewards of the MPP.

ABSTRACT. El método más simple para recolectar leche consiste en visitar a cada productor en su predio bajo la modalidad puerta-a-puerta. Así, las rutas de recolección pueden ser muy extensas, generando altos costos de transporte especialmente cuando los productores están ubicados lejos de la planta procesadora. Proponemos localizar algunos centros de recolección de leche que acumulen la producción de varios productores pequeños y distantes de la planta, facilitando el proceso de recolección. Para ello, cada centro debe disponer de una pequeña flota de vehículos que recolecte la leche de aquellos productores. Luego, camiones de mayor capacidad recolectan la leche acumulada en los centros de recolección, evitando realizar rutas extensas. Proponemos un modelo de programación lineal entera mixta utilizando AMPL-CPLEX para encontrar soluciones óptimas en instancias pequeñas que muestran la efectividad del modelo.

ABSTRACT. A previsão do prêmio dos seguros é essencial para seguradoras e resseguradoras. Ela fornece subsídios para que as seguradoras planejem suas estratégias. Apesar da importância para esse mercado, o tema da previsão de seguros é pouco difundido na literatura acadêmica. Como consequência, não há referências suficientes que indiquem o melhor tipo de modelagem para a previsão dos diversos tipos de seguro. Neste contexto, este trabalho propõe uma metodologia prática, composta por quatro etapas, que combina previsões de modelos individuais de modo a obter a maior quantidade possível de informações para prever as séries com mais acurácia. São estudadas séries de seguros de automóveis, vida, acidentes pessoais, viagem grupal, empresarial e habitacional. Os tipos de modelagem utilizados no trabalho foram: Box & Jenkins, Amortecimento Exponencial, Redes Neurais Artificiais, BATS, STL, e Naïve. Os resultados da metodologia proposta são satisfatórios, quando comparados aos dos modelos individuais.

ABSTRACT. O método GRASP vem sendo largamente utilizado no ramo de otimização e aplicado a inúmeros problemas de busca, por sua robustez e simplicidade de implementação. Uma vez que o problema de seleção de características, um dos principais e mais estudados problemas do ramo de Machine learning, é tipicamente modelado como um problema de busca, o método GRASP é naturalmente aplicável à este tipo de domínio. Este artigo apresenta um estudo avaliativo da meta-heuristica GRASP e da estratégia de convergência path relinking aplicados ao problema de seleção de características, realizando um comparativo entre estes métodos e o método de busca genética, além de realizar uma análise de sensibilidade sobre o parâmetro α do algoritmo GRASP.

ABSTRACT. Na análise envoltória de dados (DEA) é necessário, via de regra, realizar algum tipo de pré-processamento dos dados considerados no processo de análise de eficiência. Por exemplo, em muitas situações práticas, não se tem acesso à totalidade dos dados de entrada e saída referentes às unidades tomadoras de decisão (DMUs), de modo que, em tais situações, torna-se necessário dispor de alguma estratégia para lidar com esses dados faltantes. Neste contexto, o presente trabalho propõe a aplicação de uma abordagem recente de preenchimento de matrizes, conhecida como aproximação de posto reduzido, para o pré-tratamento de dados faltantes em DEA. Por meio de experimentos numéricos, a eficácia da abordagem proposta é avaliada, sobretudo no que diz respeito à proporção máxima de dados faltantes para a qual os resultados obtidos pela DEA se mantém próximos aos obtidos na situação em que não há dados faltantes.

ABSTRACT. O presente trabalho propõe um modelo que irá ajudar os tomadores de decisão a identificar os municípios mais eficientes que poderão ser candidatos a receber uma escola profissionalizante dentro de um determinado estado no Brasil. Será aplicada a Análise Envoltória de Dados (DEA - Data Envelopment Analysis) para ranquear os municípios através da avaliação de eficiência. Os indicadores que serão utilizados pela DEA foram definidos por um modelo utilizado pelo Instituto de Pesquisa e Estratégia Econômica do Ceará (IPECE). Os municípios do estado objeto do estudo serão classificados por regiões conforme os Arranjos Produtivos Locais (APL). Os resultados obtidos serão confrontados com os cenários de investimentos industriais previstos pelo Governo para cada região para validar a escolha do município-candidato.

ABSTRACT. Este articulo presenta un modelo de pronóstico multi-horizonte para series de tiempo basado en la transformada wavelet estacionaria (SWT) combinada con un modelo autoregresivo MIMO (multi-estrada/multi-salida). El modelo es implementado usando dos etapas. Primero la series de tiempo es separada en componente de alta frecuencia y baja frecuencia por el uso de SWT, la segunda un modelo autoregresivo MIMO (MIMO-AR) es calibrado y evaluado usando el metodo de los minimos cuadrados lineales. El modelo de pronostico es evaluado con series de tiempo de personas lesionadas semanalmente en accidente de transito en Santiago, Chile desde el año 2000 al 2014. La calibracion del modelo MIMOAR fue realizada con el 80% de la serie de tiempo y la fase de testing fue evaluada con el resto de los datos. El modelo de pronostico MIMO-AR propuesto para un horizonte futuro de 8 semanas durante la etapa de testing obtuvo una varianza explicada del 95% con un error absoluto medio de 0,01.

ABSTRACT. Uma abordagem contínua ao TSP é apresentada e, baseado nela, um método de direção de descida é descrito a gerar uma solução aproximada. Sendo n o número de nós, define-se um hiperplano, de dimensão n-1 e gerado por um conjunto de n-1 direções, sobre o qual se encontram as soluções viáveis. As n-1 direções são escolhidas a fim de levarem a pontos viáveis, em virtude do tamanho do passo dado, e, por isso, chamadas de direções básicas viáveis. O algoritmo tem início numa solução viável e evolui gerando novas soluções, com as tais direções básicas viáveis, e reiniciando naquela de menor valor, quando essa existir. Quando não existir melhora, busca-se no cone negativo, ou complementar, ao cone gerado por estas direções básicas viáveis, outra direção de descida. Este processo termina com o retorno ao ponto de partida, com a constatação de ótimo local, se o mesmo tiver sido gerado.

ABSTRACT. This article presents a computational tool that can be applied to minimize costs in the distribution of products or services with transport. An application of the Vehicle Routing Problem was used to optimize the routing process of deliveries of products or services. In the literature, this problem has an association with the Traveling Salesman Problem (TSP). Also in this study, the TSP is approached through a Georeferenced Routing Environment (GRE) designed by the development, implementation and testing of hybrid heuristics, built specifically for the problem. Computational experiments were performed with instances of the literature. It was possible to measure the good performance of these techniques. Heuristics are also attached and executed within the GRE, modeled with road networks maps, where it is possible create and analyze the solutions of the problem with graphic monitoring.

ABSTRACT. The accurate modeling of economies of scale in hub and spoke networks has been recently under revision. It concerns the cost structures used in hub location problems. We formulate hub location problems where the cost charged is flow-dependent, with multiple thresholds. This cost structure could be applied when leased vehicles are used to transport people or goods in hub and spoke networks. We develop efficient solution techniques and perform extensive computational experiments. Our approach is able to model economies of scale, is consistent with flow consolidation in transportation networks, and can be easily extended to other cost structures. Results show the appropriateness of our approach.

ABSTRACT. Este trabajo aborda el problema de seleccionar un proveedor externo para la gestión de residuos patógenos en una entidad universitaria. Se propone aquí el uso de una método multicriterio grupal (Procesos DRV), desarrollado para sustentar decisiones, estimular la generación de consenso y reducir el efecto de la presión grupal, a la vez que considera las perturbaciones sobre la información disponible (incertidumbre, imprecisión, datos confusos o inexistentes). Se propicia de esta manera la construcción de conocimiento compartido y el compromiso posterior con las acciones acordadas. El documento presenta y discute los resultados de una aplicación real. En las conclusiones se resalta el potencial de este enfoque metodológico para el estudio de problemas complejos de toma de decisiones.

ABSTRACT. A robustness metrics is proposed for ranking solutions in multiple criteria problems, which involves different decision makers participating in the decision process. This metrics can be understood as a flexibility measure where robustness is defined as a fitness criterion, grounded on a distance between a solution and the expected solutions representing the decision makers' preferences. Therefore, the more a solution is compatible with the decision makers' expectations, the more robust. The usage of this metrics is illustrated by the application of a multicriteria decision analysis method in an example problem.

ABSTRACT. The bullwhip-effect is an undesired upstream increasing of production variability in a supply chain, related to demand variability at the retail stage. It has extensively studied in literature. However, most of analysis use a single criterion framework. In this article, an exploratory study is developed to analyze behaviour of participants in a fictitious three-stage supply chain. The aim is to detect if they use mono-criterion or multi-criterion strategies to define the order level and control the bullwhip effect. Only trained people is considered, but two treatments (pull method and unspecific strategy) are included in the experiment. Results show that trained people fail to identify a right strategy to damp down the bullwhip effect, they are overconfident to inventory-based rules, showing resilience to multi-criteria heuristics, which are more effective in the proposed model.

ABSTRACT. In this paper, an application of the Flexible and Interactive Tradeoff (FITradeoff) method in a Colombian small company is presented. The specific decision problem is related of determining of raw materials requirements in bakery production and the construction process of decision model is detailed. The main characteristics of method and the validation of its assumptions in the problem are explained. The FITradeoff uses partial information on preferences for solving a multicriteria decision problem. Additionally, some features of a decision support system (DSS), in which the method is built in are described. The FITradeoff provides same recommendation given by the traditional tradeoff elicitation procedure, requiring less information from the decision maker and improving much more the consistence of results. Moreover, it is observed that the elicitation process is completed in fewer cycles than in classical tradeoff procedure, working with partial information.

ABSTRACT. Neste trabalho é apresentado um estudo sobre o problema de planejamento e programação da produção de bebidas à base de frutas. Trata-se de um problema de dimensionamento e sequenciamento de lotes em dois estágios de produção com particularidades: presença de um estoque intermediário entre os estágios, limpezas periódicas no primeiro e segundo estágios e a necessidade de sincronia entre os estágios. Como proposta de solução são apresentados dois modelos inteiro mistos que encontram uma solução aproximada para o problema. As soluções obtidas com os modelos passam por um pós-processamento em que é feito um ajuste da sincronia entre os dois estágios. Por serem modelos aproximados, em um a capacidade de produção é super estimada (Modelo 1 - Otimista) e no outro sub estimada (Modelo 2 - Pessimista). Testes computacionais foram realizados com instâncias baseadas em dados reais e mostram que o Modelo 1 encontra mais soluções factíveis e de melhor qualidade do que o Modelo 2.

ABSTRACT. Assembly lines are a widespread production method used to manufacture goods in high volumes. Line balancing deals with one of the problems that need to be tackled in order to manage the assembly line; that is, the assignment of operations that need to be performed to the final product to the stations that constitute the assembly line. In this work we consider a category of line balancing problems in which the objective to jointly assign tasks and tools to the workstations. In order to solve the problem, lower and upper bounds are proposed and the quality of the proposed methods are analyzed.

ABSTRACT. In this work we use alpha-point scheduling techniques to compute approximate solutions for a general family of scheduling problemas in a job shop setting: each production order (PO) has several jobs, with precedence constraints between them, and the goal is to compute a schedule that minimizes the sum of the makespan of all the POs. Our numerical experiments, on real data of a job shop setting, show that the approximation ratio is small even when resource constraints are added.

ABSTRACT. Makespan scheduling on identical machines is one of the most fundamental packing problem in the discrete optimization literature. It asks for an assignment of n jobs to m identical machines that minimizes the makespan. Recently, Chen et al. showed that the running time of a (1+ε)-approximation algorithm must have an exponential dependency on 1/ε, assuming the Exponential Time Hypothesis (ETH). Many algorithms exist that try to obtain low dependencies on 1/ε, however they all leave a gap with respect to mentioned lower bound. Our main result is an algorithm with an (almost) tight running time under ETH. This is achieved via a new structural insight on the configuration-IP, namely, the existence of a highly symmetric and sparse optimal solution. This can then be exploited by integer programming techniques and enumeration. We also show how to extend our algorithm to the related machines setting and to a larger class of objective functions on parallel machines.

ABSTRACT. En este paper estudiamos el Nurse Rostering Problem (NRP) considerando un horizonte de planificación de 28 días, tres turnos diarios y tres categorías de enfermeras. Nosotros formulamos el problema como un Integer Linear Problem (ILP) para determinar la óptima asignación de enfermeras que minimiza la insatisfacción de los requerimientos de estas, tanto de asignaciones deseadas como de turnos libres. Para las instancias que probamos, el problema es tratable con tiempo promedio de 20 segundos. La factibilidad del problema radica en tener una cantidad apropiada (y mínima) de enfermeras dispuestas a trabajar en turnos de noche, las cuales a petición del manager pueden cubrir a lo más un turno de día cada dos semanas.

ABSTRACT. En este estudio se presenta una formulación matemática basada en programación entera-mixta que representa el problema de asignación de turnos de enfermería en un hospital. Se considera un conjunto de enfermeras y un conjunto de turnos de enfermería definidos: Mañana, tarde, Noche y descanso, Se asume que la planeación se hace para un horizonte de tiempo discreto y finito. El modelo decide la planeación de horarios semanales para cada una de las enfermeras considerando restricciones de tipo laboral e institucional. El modelo propuesto considera dos conjuntos de restricciones: “Hard-constrains” y “soft-constraints” las cuales son penalizadas en la única función objetivo con un factor constante de “castigo”. Se compara el desempeño de un método exacto contra una heurística constructiva en un conjunto de 75 instancias aleatorias de diferentes tamaños. Los resultados obtenidos de la comparación entre el modelo matemático y el método heurístico sugieren un excelente desempeño de este último

ABSTRACT. O Problema de Escalonamento de Médicos (PEM) pode ser definido como a alocação de tarefas a grupo de médicos durante um horizonte de planejamento, considerando um conjunto de regulamentações de trabalho, preferências individuais e regras de organização, regras estas que são muitas vezes conflitantes. O problema lida majoritariamente com objetivos qualitativos, sendo neste caso, produzir escalas de modo a maximizar o atendimento das restri ̧c ̃oes de preferência pessoal, respeitando as restrições laborais e organizacionais. Neste artigo apresenta-se um modelo de Programação Inteira Mista fundamentado em obras da literatura e propõe-se uma heurística derivadas da formulação matemática para resolução eficiente de instâncias baseadas em dados reais do PEM.

ABSTRACT. La planeación efectiva de capacidad en condiciones de incertidumbre garantiza planes robustos y consistentes en el tiempo. El sector Salud, es un sistema de servicio de gran relevancia para la aplicación de métodos que consideren la inter-temporalidad en las decisiones, dado que una planeación deficiente afecta directamente el bienestar de las personas. Este trabajo busca construir una modelación de decisiones jerárquicas en múltiples etapas, aplicado a la planificación de camas en el sector Salud. Para ello proponemos una metodología de solución basada en la formulación y resolución de modelos de optimización estocástica. El problema consiste en determinar la disponibilidad de camas para un hospital en el sistema de Salud pública que minimice una función del tiempo de espera de los pacientes. A partir de la solución de los modelos de optimización, se definen políticas de planificación que permitirán tomar mejores decisiones en distintos horizontes de planificación.

ABSTRACT. This work studies the problem of maximizing the lifetime in a general class of wireless sensor networks. Two kinds of sensor devices are considered: wireless directional sensors networks and sensors able to modify their sensing ranges. In both cases it is considered that information collected by the sensors must be addressed to a single base station via multi-hop wireless communication. The problem is formulated through a large scale linear model and solved via column generation. The proposed approach decomposes the problem in a pricing subproblem deciding the role each sensor play at a given time, i.e. the orientation or the sensing range, and time-related decisions indicating the total time that a given subset of sensors operate. A new model for the pricing subproblem is proposed and a cutting planes approach is used to solve it. Preliminary results seem to indicate that the proposed approach is efficient over medium-large size instances.

ABSTRACT. En el Municipio de Piedecuesta, las problemáticas sociales son abordadas en el Centro de Proyección Social CPS. Las problemáticas sociales identificadas están en tres ámbitos de las competencias ciudadanas: La disociación social impide llegar a consensos comunitarios, (convivencia y paz); Problemas de negociación dentro de los estamentos sociales, (Pluralidad, identidad y valoración de las diferencias); Falta de reconocimiento de la ley, el respeto por la democracia y los valores cívicos, (Participación y responsabilidad democrática); Dificultad para identificar problemas en áreas: social, cultural, sexual y educativo, (convivencia y paz). En este contexto, es necesario identificar estrategias orientadas a potenciar competencias ciudadanas, a nivel cognitivo, a partir del Proceso de Jerarquía Analítica. En esta ponencia se presenta los resultados de la aplicación PJA en la selección de estrategias, a partir de las respuestas expertos, ubicados en 7 colegios del Municipio.

ABSTRACT. Stackelberg Security Games (SSGs) are a class of games where one defender (leader) wants to protect certain areas or targets committing to a mixed strategy, and then one (or many) attacker (follower) observes this strategy over time and chooses where to attack. In this work we focus on a SSG defined on a graph generated by adjacent clusters of targets (called areas). Due to limited resources the patrolling strategy requires that areas must be paired and a joint patrol is deployed to protect a target within one of these areas. This creates a game with a defender strategy space which is exponential in the number of possible targets. We propose a new formulation where the decision variables are coverage frequencies on the edges of this graph and on the location set dramatically reducing the size of the defender strategy space, which allows the solution of this problem for real instances. To build implementable strategies from this solution we propose and compare two sampling methods.

ABSTRACT. In recent years, the traditional method to analyze and compare mental models has been extended to include feedback loops, which are important components for situations with endogenous dynamics. The extended “distance ratio” method takes into account feedback loops recognized by decision-makers. However, the causal structure of articulated mental models of dynamic systems contains many more loops than are typically recognized by decision-makers. A trained analyst would recognize the shortest independent loop set. This paper uses data of an ongoing research project and applies the established method to both loop sets. It thereby shows how much the “model distance ratio” is impacted by the choice of which feedback loops are taken into account. It is argued that the comparison of model distance ratios between these two types of loop sets up the possibility to investigate the relationship between performance and dynamic decision-making experiments and systemic reasoning capabilities.

ABSTRACT. Determinar el efecto en el empaquetamiento rectangular teniendo como parámetros de control diferentes objetos (Círculos, cuadrados, diamantes, octágonos) y dos familias de desigualdades validas. Usando métodos de solución exacta por medio de análisis estadístico, por cada tipo de objeto se tienen entre 12 a 14 tamaños diferentes del mismo objeto. En el embalaje sin permitir la telescopia se noto que en los primeros 240 minutos de ejecución del solver (30,60,120 y 240), tanto las desigualdades validas usadas como los diferentes objetos no presentan significancia estadistica al 95 %, para la variable respuesta GAP (valor P>0.05). Dicho fenómeno no es similar a los 480 y 600 minutos en donde el tipo de objeto si afecta signicativamente al GAP (valor P<0.05), pudiéndose dar este patrón principalmente al crecimiento combinatorio del problema en esos instantes.

ABSTRACT. In this presentation, we focus on the covering problem of a three dimensional body using different radius spheres, aiming its use on the treatment planning of Gamma Knife radiosurgery.Gamma Knife unit delivers accurately a suitable dose of ionizing radiation, called shots,to the target tumor region. In good approximation, these shots can be modeled as spheres of different sizes. Multiple shots can be used to cover the entire tumor, while avoiding an excessive dose to the surrounding healthy tissue. We approach the problem just by the geometric covering point of view, that is, given a set of spheres and a body, the objective is to fully cover the body using the smallest possible number of spheres, regardless of the dosage issue. In order to solve this mathematical programming problem, we consider an approach based on the application of penalty and stochastic heuristic search techniques.

ABSTRACT. In this paper, we investigate conditions for cargo static stability in packing problems. In particular, we address the constrained two-dimensional knapsack problem, which consists in orthogonally packing rectangular items in a single rectangular container, with the aim of maximizing the packed area. An integer linear programming model that positions the items over a discrete grid of points is used to solve the problem. The model has embedded constraints that ensure a minimum support for the base of each item. The solutions obtained with the model for different support factors and grids of points are then evaluated by a procedure that verifies the conditions for the static stability of the resultant packing. Computational tests were performed with this approach and a large variety of randomly generated instances, and the results show that some classes of instances are more likely to present static instability than others.

ABSTRACT. A caterpillar is a graph such that the removal of all its vertices with degree 1 results in a path. Given a graph G, a caterpillar-packing of G is a set of disjoint (not necessarily induced) subgraphs of G such that each subgraph is a caterpillar. In this work we consider the set of caterpillar-packings of a graph, motivated by the fact that such structures correspond to feasible solutions of the 2-schemes strip cutting problem with a sequencing constraint (2-SSCPsc) presented by F. Rinaldi and A. Franz in 2007. We study the polytope associated with a natural integer programming formulation of this problem, which is called the "caterpillar-packing polytope". A key feature of this polytope is the existence of straightforward facet-preserving procedures, i.e., procedures that take as input a facet-inducing inequality and generate slightly modified inequalities that also are, under certain hypotheses, facet-inducing. In this work we present new such procedures for this polytope, and explore separation procedures based on these ideas.

ABSTRACT. This paper addresses the economic lot-sizing problem with production and remanufacturing in which the returns are sorted into different levels of quality. We provide a mathematical formulation for the problem and show that it can be as hard as solve that the problem with ungraded returns. Different inventory policies are suggested and compared by means of adding specific constraints to the original formulation. Useful managerial insights are obtained from the numerical experiment carried out, such as it is suitable to either produce or remanufacture but not both in the same period and that it is profitable to remanufacture those returns of only one quality class which in turn may be not the higher one.

ABSTRACT. Estudiamos un sistema con un conjunto de puntos de venta geográficamente dispersos. Se asume un comportamiento aleatorio de la demanda. Además, se considera un sistema mono-producto, la administración de la política de inventarios con un único proveedor y el traspaso de productos entre instalaciones. Se plantea un modelo estocástico multi-etapa, con el objetivo de maximizar el beneficio obtenido por el minorista al final de un horizonte de planeación determinado. Se propone usar programación estocástica dinámica para el problema de un único retail. Por otra parte, proponemos usar Programación Estocástica Dinámica Dual y Programación Dinámica Aproximada para el problema con múltiples puntos de venta. Se presenta un análisis de capacidades de cada método. Los tiempos computacionales para el problema muestran una dependencia directa del número de estados posible, así como del número de periodos y la cantidad de posibles decisiones en la variable del control del sistema (cantidad ordenada)

ABSTRACT. En este trabajo se estudia un modelo del tamaño del lote para artículos cuya demanda varía con el tiempo y es sensible al precio de venta. Se considera que el coste unitario de almacenamiento es una función no lineal dependiente del tiempo que el artículo permanece en el inventario. Se admite la posibilidad de escasez en el inventario. En ese caso, solamente una fracción variable de la demanda se sirve con retraso. Se formula el problema de inventario y se presenta un procedimiento algorítmico para determinar la política óptima del inventario. El sistema aquí desarrollado incluye como casos particulares varios modelos de inventarios EOQ estudiados por otros autores. Finalmente, los resultados se ilustran con varios ejemplos numéricos.

ABSTRACT. Este trabajo estudia el problema de abastecimiento coordinado entre una bodega y varios minoristas (N) bajo un ambiente de demanda estocástica. El objetivo es encontrar la política que minimiza el costo total, evitando sacrificios económicos relacionados con división del problema en N modelos independientes. Para resolver este problema en su versión deterministica, se han planteado políticas de ciclo único o compartido. En este tipo de políticas todos los minoristas tienen inventario igual a cero cada vez que la bodega es reaprovisionada. Abdul-Jalbar et. al (2010), propone una solución a dicho problema en formulaciones deterministicas. Nuestro trabajo considera una demanda estocástica bajo una función de costos S,T. Dicha técnica se basa en una extensión estocástica de la heurística de Abdul-Jalbar (2010).

ABSTRACT. In this paper we solve a practical problem arising in the ATM management and replenish ments in Santiago, Chile. This rich and challenging problem shares much of its structure with the inventory-routing problem, but some features makes it unique. We model the problem and solved it by branch-and-cut. In order to efficiently tackle large instances, we propose a new Variable MIP Neighborhood Search (VMNS) heuristic, which takes advantage of the MIP implementation. The VMNS relies on the mathematical formula tion of the problem and significantly simplifies its resolution. We derive neighborhoods to exploit the structure of the problem at hand, be it over routes, locations, periods or quantities delivered. Based on extensive computational experiments, our VMNS is shown to significantly improve the solutions from the exact method using only a fraction of the time. Sensitivity analyses are performed to confirm the robustness and effectiveness of our method.

ABSTRACT. El problema de planificación de cosechas se ha estudiado desde muy diversos puntos de vista (maximización del beneficio, máximo rendimiento del agua disponible, sostenibilidad del plan de cosecha…) En este trabajo se aborda un nuevo enfoque considerando las nuevas formas de relación proveedor-cliente, en las que priman los acuerdos de colaboración a largo plazo, y donde las demandas vienen establecidas con antelación. Esto da lugar a que las restricciones deban garantizar unos determinados niveles de producción en ciertas fechas. Se presenta aquí un modelo biobjetivo, de tipo lexicográfico, en el que además de minimizar costes, se pretende minimizar el riesgo de no atender las demandas, usando dispersión geográfica. Se presentan unos resultados ilustrativos para mostrar la efectividad del modelo.

ABSTRACT. El trabajo presenta un resumen de los principales resultados del desarrollo de una plataforma de apoyo a la toma de decisiones para la agroindustria presente en la zona central de Chile. Se presentan aplicaciones WEB que tienen por objetivo apoyar a tres importantes sectores de la agroindustria (aceite de oliva, deshidratado y jugos) en tres actividades claves de su proceso. Las actividades son: la planificación de cosecha en el sector aceite de oliva, la gestión del patio de acopio en el sector de jugos y la planificación de corto plazo en el sector de deshidratados. Se presenta la importancia de los problemas seleccionados, los modelos matemáticos con los cuales fueron abordados, las estrategias utilizadas para dar solución a los modelos y una cuantificación del impacto esperado de su uso. Adicionalmente, se entregará un conjunto de problemas asociados de interés empresarial y que representan desafíos científicos para el área de investigación de operaciones.

ABSTRACT. One of the main factors that affect the fresh apple quality is the ripening point at which the fruit is harvested. For this reason, harvest planning should consider this factor and schedule the efficient use of resources when the fruit is in its picking time. The main resources used during the apple harvest are labor, equipment and machinery, where labor is the more scarce and critical resource. In this sense, if the number of harvester is not enough to pick the fruit when is mature, the quality of fruit could be damaged. This situation become more complicated when labor is shared by several orchards belonging to a same company. This is a common situation in Chilean companies. In order to schedule and assign harvest resources in a better way during a season, an optimization model for supporting planning decisions of several orchards with shared resources is developed. The proposed model was applied to a case study considering three apple orchards.

ABSTRACT. Processing plants are central for the operation of fruit supply chains. One of the main aspects to consider is fruit transportation to the processing plant. Hence, this work proposes a two-stage stochastic linear programming model to support the fruit transport planning from the storage facilities to the processing plant. The aim of the model is to minimize the daily transportation costs and associated costs of different storage facilities from where fruits are supplied to the plant in order to meet the demand. The model considers plant processing capacity, fruit demand, number and type of trucks available and the inventory of fruit in each type of storage facilities. Finally, the model was applied to a real case study of a processing plant located in the O'Higgins Region (Chile), where reported savings only in transport costs reached about 16 percent.

ABSTRACT. Forest planning is not an easy task due the influence of climate change that impacts the forest growth, occurrence of wildfire and other natural disturbances. In climatic predictions, the real data is characterized by uncertainty. In this paper we address the problem of harvest planning considering uncertainty of climate change, which affects tree growth, including spatial constraints. We develop a solution approach for this problem for a forest located in Portugal. We consider 32 climatic scenarios, each one associated to a tree growth scenario. We develop a MIP formulation where the stochastic problem represents the forest planning problem, adding the non-anticipativity constraints that link all scenarios. The resulting problem is called the extensive form. This form can lead to a very large problem that can only be solved by decomposition into scenarios. We solve the problem using the Progressive Hedging (PH) algorithm, which decomposes the problem into scenarios.

ABSTRACT. Nowadays, social interaction plays an important role in life, allowing people to achieve their desires, like a product you want to buy, a new professional contact or the resumption of a friendship. In this context, it is possible to receive recommendations that will help fulfill social goals, improving not only our experience as users but also the experience of others. We present a recommendation technique between items based on a mixed approach of collaborative and content filtering, and social network analysis. From a root item, a relationship's graph is generated and used to extract network metrics. The similarity of the root item with other items present in the graph is evaluated and combined with the obtained metrics. We test these contributions in a usecase, a co-authorship network. The combination of the similarity with metrics for social network analysis of the relationship's subgraph allows the configuration of recommendation algorithms according to different semantics of use.

ABSTRACT. In this study, we propose the use of simultaneous Xbar charts to control autocorrelated processes. The first set of Xbar charts is side-sensitive with regard to the same variable (SV Xbar charts) and the second one is side-sensitive with regard to both variables (BV Xbar charts). In order to reduce the negative effect of the autocorrelation on the charts’ performance we build up the samples taking one item form the production line and skipping one, two, or more before selecting the next one. In comparison with the standard synthetic T2 chart, the SV and the BV charts signal faster in a wide variety of disturbances, except when the variables are high correlated. The BV charts are simpler and signal faster than the SV charts.

ABSTRACT. One of the most common methods used with social network analysis of criminal groups is node importance evaluation, which focuses on the links between network members to identify likely crime suspects. This study introduces a new approach that incorporates members’ individual criminal propensities. Since that the traditional node evaluators dont take advantage of this additional information, a new evaluator is proposed. The proposed evaluator employs a novel perspective based on concepts of human and social capital, a structure built on ego networks and an analogy between social interaction and field theory. This evaluator is applied on a case to test its effectiveness by comparing its performance with that of traditional evaluators. The results show that the proposed evaluator performs better than the others and that the integration of the criminal propensity factor into a criminal network enables the evaluator to identify key criminal suspects more accurately.

ABSTRACT. In this paper, we introduce a methodology based on zero-inflated long-term survival data in order to deal with fraud rate estimation in bank loan portfolios. Our approach enables us to accommodate three different types of loan borrowers, i.e., fraudsters, those who are susceptible to default and finally, those who are not susceptible to default. Regarding to the survival analysis framework, an advantage of our approach is to accommodate zero-inflated times, which is not possible in the standard cure rate model introduced by Berkson & Gage (1952). The parameter estimation is reached by maximum likelihood estimation procedure. To illustrate the proposed method, a real dataset of loan survival times is fitted by the zero-inflated Weibull cure rate model.

ABSTRACT. Detecting malicious emails is a huge challenge and at the same time very relevant, especially in times of an increasing threat by cybercrime. The interaction between sender and receiver of such messages can be modeled as a game between two players. In adversarial classification, it is natural to model such an interaction as a dynamic game of incomplete information, since the classifier (receiver) does not know the exact intentions of the different types of adversaries (senders). For these games, equilibrium strategies can be approximated and used as input for classification models. We show how to model such interactions between players. We propose perceptron-like machine learning approximations as well as novel Adversary-Aware Online Support Vector Machines. Results in a real-world adversarial environment show that our approach is competitive with benchmark online learning algorithms, and provides important insights into the complex relations among players.

ABSTRACT. La distribución de energía eléctrica es un sistema que requiere alta disponibilidad ya que el 90%de las fallas afectan la continuidad del servicio. Todo sistema es susceptible de mejorar y dentro de las formas que podrían aplicarse están: Protecciones especiales, automatización del sistema, respuesta más rápida de las cuadrillas de mantenimiento, mantenimientos preventivos programados, etc. Actualmente, una falla es reportada al centro de distribución donde se genera una orden de servicio, que es asignada a la cuadrilla más cercana , sin embargo, la asignación no es eficiente ni uniforme, , lo cual es una debilidad porque no se están optimizando los recursos, necesarios para mejorar la atención. En este artículo se propone, mediante una técnica de ruteamiento de vehículos como el VRP , la cual se solucionará usando una heurística mediante un modelo MTZ para asignar a las cuadrillas las fallas que deben atender, de manera que se optimicen el tiempo o la distancia a recorrer.

ABSTRACT. Um modo eficiente para se determinar o estado de um sistema de energia elétrica é através do problema de Fluxo de Potência Ótimo. O problema de Fluxo de Potência Ótimo é modelado como um problema de programação não linear com variáveis discretas e contínuas. Devido à dificuldade de resolução causada pelas variáveis discretas em problemas de programação não linear, e a importância de se resolver este problema de forma eficiente, propõe-se neste trabalho uma nova abordagem de resolução para este problema. Uma função senoidal para minimização do gap de integralidade das variáveis é proposta, e uma sequência de problemas auxiliares com somente variáveis contínuas é obtida. Os problemas auxiliares são resolvidos por um método de pontos interiores. As soluções dos problemas auxiliares convergem para a solução do problema original com variáveis discretas e contínuas. Testes numéricos com os sistemas elétricos IEEE demonstram a eficiência da abordagem proposta.

ABSTRACT. Many types of models are used for evaluating climate-change-related programs and policies, ranging from “top down” models of the entire economy (e.g., computable general equilibrium, or CGE models) to specialized models focused on specific markets (e.g., Oak Ridge’s MA3T model of the US personal vehicle market). Somewhere in the middle are “bottom up” models of the energy sector (so-called 4E, or “energy/economy/environment/engineering” models). A major concern is that 4E models, despite having a high level of technological detail, yield unrealistic consumer market responses, calling into question their value for making policy decisions. We have developed a framework based on sound economic theory that allows 4E models to be modified to produce more realistic market behavior. The approach is demonstrated for personal vehicle purchases using a widely used optimization-based modeling system (TIMES/MARKAL), extending a previously developed model of California’s energy sector (CA-TIMES).

ABSTRACT. El reposicionamiento de bicicletas en los sistemas de bicicletas públicas es un problema que incide fuertemente en la cantidad de recursos necesarios para su operación y limita la capacidad financiera de los mismos, por lo que ha dado lugar a que en últimos años, hayan surgido numerosas formulaciones bajo los enfoques estático y dinámico. Por la rotación de bicicletas en el sistema, es necesario contar con una estrategia particular para ciertos períodos del día, mismos que son determinados de manera conjunta con el análisis dinámico de la demanda. En este trabajo se presenta una formulación bi-objetivo, donde se modela la estrategia de redistribución de bicicletas en las estaciones de un sistema de bicicletas públicas en una ciudad en México, a partir de los criterios de la reducción de los costos involucrados en el reposicionamiento y en el cumplimiento de la satisfacción de la demanda. El modelo, por su complejidad se resuelve con una estrategia evolutiva basada en el NSGA-II.

ABSTRACT. Soybean and corn production has increased steadily in Brazil and Paraná State is the second largest producer of these grains in the country. The increased production has now necessitated to increase the storing facility also. Partitioning the storage is a proposal to aggregate the municipalities of Paraná into regions as a way of facilitating production and transportation of the grains. This paper aims to organize the storage regions of Paraná by modeling its municipalities as a multi-objective graph (territory) partitioning problem with the municipalities being the nodes and roads linking them as the edges of the graph. In order to find the effective number of new silos to be constructed and region-wise their locations, maximization of the homogeneity of storage deficit and minimization of the distances from production sources to storage points are considered as two objective functions of the problem.

ABSTRACT. Varias ciudades en el mundo han decidido peatonalizar ciertas zonas de la ciudad, principalmente el centro. Al respecto distintos autores han analizado los efectos de dicha medida y desarrollado tecnologías para hacer mediciones tanto del flujo peatonal como vehicular. Sin embargo, este artículo orienta su análisis, en desarrollar un modelo de optimización multiobjetivo para generar una propuesta que permita determinar la porción de área a asignar para uso de flujo peatonal y para uso de flujo vehicular en una determinada zona. El estudio valida su análisis para un kilómetro cuadrado del centro de la ciudad de Bogotá y lo hace mediante técnicas de optimización bioinspirada; particularmente los algoritmos de NSGA II, NSPSO y BFOA. Luego, usa TOPSIS como técnica para seleccionar la mejor alternativa posible dado los resultados obtenidos con dichos algoritmos.

ABSTRACT. The Urban Transit Routing Problem (UTRP) is a NP-hard combinatorial optimization problem that consists of finding a set of routes for a transportation system maximizing the comfort of passengers and minimizing the operational costs. In this work we propose a Multiobjective island model genetic algorithm that works with a set of populations. Each population undergoes its own genetic phases: selection, transformation and elitism. Each population posses its own weighing of problem objectives focusing its search on special zones of the Pareto front. Information is exchanged between populations by migration of solutions. We evaluated our approach using well-known UTRP benchmarks and compared it with an equivalent genetic algorithm and some local search approaches from literature. Our approach shows good performance finding better quality solutions in the central zones of Pareto fronts. For future work we plan to study its genetic operators to improve its search on border zones of the front.

ABSTRACT. En esta investigación se presenta un problema multiobjetivo de diseño de cadena de suministro de tres escalones, incorporando decisiones tácticas y estratégicas con múltiples minoristas y grupos de potenciales de plantas y almacenes. La principal contribución del estudio es extender la literatura actual incorporando el análisis de la técnica justo a tiempo JIT en un problema de localización-inventario considerando una demanda estocástica, política de inventarios y decisiones de localización y asignación. Generando un diseño de cadena de suministro que satisface la minimización de costos de operación de la cadena de suministro y la minimización de la cantidad de la orden bajo el enfoque JIT. La política de inventarios en los costos de operación y cantidad de la orden JIT genera un modelo no lineal. Para esto se propone una aproximación lineal a tramos, generando una formulación no lineal entero mixto que es resuelto con un solver comercial utilizando el método de epsilon-restricción.

ABSTRACT. This work is motivated by the production scheduling problem on a flexible machine faced by a metal working industry. We are given a certain number of products. Each product has a due date which is a period in a planning horizon, and is composed by a number of jobs. The processing of each job requires certain tools. The machine is assisted by a CAM software which allows automatic changing between two different jobs given that the tools required to processing are loaded. Because the machine has a limited capacity for the storage of tools, it has to be stopped when a job requires at least one tool that is not already loaded. Thus, the problem consists of deciding the jobs to be processed in each day of the horizon, along with their processing order, to minimize overtime and weighted tardiness costs. We propose an integer programming model and heuristics for the problem. We present numerical results on real data, and significant gains were obtained with respect to the firm's practice.

ABSTRACT. Este trabalho aborda o Problema de Estoque e Roteamento de Múltiplos Veículos onde o fornecedor distribui, usando uma frota homogênea de veículos, um único tipo de produto ao longo de um horizonte de planejamento finito. O principal objetivo é a minimização dos custos operacionais de transporte e armazenamento dos produtos. Para resolver o problema é proposto um algoritmo baseado na metaheurística Iterated Local Search (ILS), do qual incorpora uma heurística Variable Neighborhood Descent com ordenação randômica de vizinhanças (RVND) durante a etapa de busca local. Além disso, um procedimento exato baseado em programação matemática é combinado ao algoritmo desenvolvido com a finalidade de resolver, de forma ótima, o subproblema de gerenciamento de estoque. Para avaliar a abordagem desenvolvida, testes computacionais são realizados em 560 problemas-teste, alcançando resultados competitivos em comparação aos melhores algoritmos da literatura.

ABSTRACT. In this paper, we propose a grasp metaheuristic for the longest link node deployment problem (LLNDP). The llndp is NP-hard and models a machine assignment problem related to the optimal usage of resources in a cloud computing environment. In a previous work [10], greedy and randomized heuristics were proposed for the llndp. Computational tests carried out here on 450 instances of various sizes and from three dierent classes revealed that our grasp outperformed those heuristics with an average gain of 14.2% in the solution cost. An in-depth discussion of the results is presented.

ABSTRACT. Time-to-target plots (tttplots) or runtime distributions are a useful tool to characterize, evaluate, and compare the behavior of randomized algorithms. However, they are limited to the evaluation of one single problem instance at-a-time. In this work, we propose their extension to address sets of test problems with multiple instances. Numerical results for different problems illustrate the applicability and usefulness of the newly proposed m-tttplots tool.

ABSTRACT. The surgical suite has multiple and powerful stakeholders. In a public hospital, the government wants to achieve some social measures like: number of patients in the waiting list, number of days in the waiting list, or percentage of patients treated after the clinically acceptable period. The hospital administration wants to achieve those goals in order to avoid high contractual penalties; they also desire a high efficiency level of the surgical suite. However, surgeons often schedule surgeries according to their agenda and their capacity to remember all their patients. A systematic system to select and schedule the patients to be operated in a given week is under development, trying to consider the different stakeholders’ perspectives for a surgical case assignment problem. In this work we present the Integer Linear Programming models considered and the heuristics that were already developed. The results, using data from a Portuguese hospital, will be compared and discussed.

ABSTRACT. En esta investigación se caracteriza la eficiencia en el uso de los recursos al adoptar reglas de despacho simples, en el proceso de programación de quirófanos, en una institución de salud regional. Se considera la limitada disponibilidad de especialistas para llevar a cabo los procedimientos quirúrgicos, principal restricción del sistema, que condiciona la programación a franjas horarias predeterminadas por dicho especialista y las condiciones de operación y demanda de la clínica objeto de estudio. El propósito del estudio es determinar las combinaciones más apropiadas de reglas de despacho y esquemas de programación, para mejorar el flujo de pacientes, el uso de las salas y la oportunidad de atención, de una institución de salud con limitada disponibilidad de recursos y difícil acceso al uso de software especializado para la programación de los quirófanos.

ABSTRACT. Se describe un método cuantitativo para resolver el problema del establecimiento de prioridades en la programación de cirugías para listas de espera de usuarios de servicios de cirugía en Instituciones Prestadoras de Servicios de Salud, basado en el impacto que tienen los resultados de las intervenciones en la mejora de la calidad de vida del paciente independiente del componente financiero. Esto facilitará la formulación de modelos de optimización donde el equilibrio de múltiples objetivos (p.ej económicos y sociales) puede formularse en una misma escala de medida. El modelo establece una variable no observable –latente- denominada “Calidad Asistencial” como eje para diseñar la escala donde los usuarios serán medidos con base en información proveniente de valoraciones que normalmente se hacen por medio de métodos cualitativos. La variable se convierte a escala de intervalo con un modelo de elección discreta, diseñado con las propiedades de la función Logit.

ABSTRACT. La programación de las Salas de operaciones es una tarea de vital importancia, debido a que este recurso es el más costoso en un hospital. Sin embargo, diferentes fuentes de incertidumbre hacen compleja su programación. El principal desafío consiste en manejar la duración incierta de las cirugías, la que puede generar: retrasos, horas extras, llegando incluso a la cancelación de cirugías. Por esto, es necesario desarrollar herramientas de apoyo que protejan contra la incertidumbre la programación de las Salas de Operaciones. Nosotros, estudiamos el problema de la programación de cirugías minimizando los costos de retrasos, tiempo extra y la cancelación de cirugías. Para resolver, proponemos dos enfoques basados en modelos de optimización robusta. También, desarrollamos métodos alternativos para resolver los enfoques.

ABSTRACT. In 2000, Chile introduced a profound Health Reform (AUGE), aiming at a more equitable and fair system. For a list of diseases, it sets a service level which includes a maximum waiting time between diagnosis and treatment. If this is exceeded, the patient is treated in the private health system, with the corresponding higher expenses. This creates an important management problem for the hospital, for which is now imperative to schedule the surgeries including the additional costs set by the AUGE Plan. With the collaboration of the Instituto de Neurocirugía, we developed a mathematical model to schedule surgeries in order to minimize the additional cost of referrals to the private sector, complying the AUGE time guarantees, and constraints regarding operating rooms and ICU bed capacities. We develop a tactical and operational model which has been tested using date collected from the hospital in the last 18 months.