STEPS TO IMPLEMENT
BAYESIAN INPUT DISTRIBUTION SELECTION 


Stephen E.
Chick Department of Industrial and Operations Engineering The University of Michigan 1205 Beal Avenue Ann Arbor, Michigan 481092117, U.S.A. 

ABSTRACT  
There are known pragmatic and theoretical difficulties associated with some standard approaches for input distribution selection for discreteevent simulations. One difficulty is a systematic underestimate of the variance of the expected simulation output that comes from not knowing the `true' parameter values. Another is a lack of quantification of the probability that a given distribution is best. Bayesian methods have been proposed as an alternative, but acceptance has not yet been achieved, in part because of increased computational demands, as well as challenges posed by the specification of prior distributions. In this paper, we show that responses to questions like those already asked and answered in practice can be used to develop prior distributions for a wide class of models. Further, we illustrate techniques for addressing some computational difficulties thought to be associated with the implementation of Bayesian methodology.  
Return to
Wintersim 
USING INPUT PROCESS
INDICATORS FOR DYNAMIC DECISION MAKING 


Michael
Freimer School of Operations Research and Industrial Engineering Cornell University Ithaca, NY 14853, U.S.A 
Lee
Schruben Department of Industrial Engineering and Operations Research University of California at Berkeley 4135 Etcheverry Hall Berkeley, CA 947201777, U.S.A  
ABSTRACT  
In a continually changing environment, a simulation study that integrates the activities of data collection, model analysis, and decision making has some distinct advantages. In this paper we look at simulation project dynamics from a highlevel and examine some ways for integrating these activities. From this perspective, the processes used to drive a simulation model are forecasts of environmental changes, and the parameters for models of these processes are viewed as leading indicators. A simple decisionmaking scenario having some of the characteristics of semiconductor manufacturing is used to illustrate the ideas.  
Return to
Wintersim 
REGRESSION METAMODELLING
IN SIMULATION USING BAYESIAN METHODS 


Russell C. H.
Cheng Faculty of Mathematical Studies University of Southampton Highfield Southampton, SO17 1BJ England 

ABSTRACT  
This paper further develops some of the ideas set out by Cheng (1998) for output analysis using Bayesian Markov Chain Monte Carlo (MCMC) techniques, when a regression metamodel is to be fitted to simulation output. The particular situation addressed by Cheng was where there is uncertainty about the number of parameters needed to specify a model. This arises because there may be uncertainty about the number of terms to be included in the regression model to be fitted. The statistically nonstandard nature of the problem means that it requires special handling. In this paper we shall use the derived chain method suggested by Cheng (1998). However, whereas in that paper the distribution of the response output of interest was assumed to be simply normal, it is typically the case, especially in the study of systems working near their capacity limit, that this distribution is skewed, and moreover the distribution has a support that is effectively bounded below  that is the distribution has a threshold. We describe how the derived MCMC method might be applied in this situation and illustrate with a numerical example involving the simulation of a computer PAD network.  
Return to
Wintersim 
VARIANCE REDUCTION OF
MONTE CARLO AND RANDOMIZED QUASIMONTE CARLO ESTIMATORS FOR STOCHASTIC
VOLATILITY MODELS IN FINANCE 


Hatem Ben
Ameur Service de l' Enseignement des Methodes Quantitatives de Gestion École des Hautes Etudes Commerciales 3000, chemin de la CoteSteCatherine Montréal, H3T 2A7, CANADA 
Pierre
L'Ecuyer Christiane Lemieux Département d'Informatique et de Recherche Opérationnelle Université de Montréal, C.P. 6128, Succ. CentreVille Montréal, H3C 3J7, CANADA  
ABSTRACT  
We illustrate by numerical examples how certain variance reduction methods dramatically improve the efficiency of Monte Carlo simulation for option pricing and other estimation problems in finance, in the context of a geometric Brownian motion model with stochastic volatitity. We consider lookback options and partial hedging strategies, with different models for the volatility process. For variance reduction, we use control variates, antithetic variates, conditional Monte Carlo, and randomized lattice rules coupled with a Brownian bridge technique that reduces the effective dimension of the problem. In some of our examples, the variance is reduced by a factor of more than 100 millions without increasing the work. The examples also illustrate how randomized quasiMonte Carlo can be effective even if the problems considered involve a large number of dimensions.  
Return to
Wintersim 
EFFICIENCY IMPROVEMENTS
FOR PRICING AMERICAN OPTIONS WITH A STOCHASTIC MESH



Athanassios N.
Avramidis Paul Hyden School of Operations Research and Industrial Engineering Cornell University Ithaca, NY 14853, U.S.A. 

ABSTRACT  
We develop and study generalpurpose techniques for improving the efficiency of the stochastic mesh method that was recently developed for pricing American options via Monte Carlo simulation. First, we develop a meshbased, biasedlow estimator. By recursively averaging the low and high estimators at each stage, we obtain a significantly more accurate point estimator at each of the mesh points. Second, we adapt the importance sampling ideas for simulation of European pathdependent options in Glasserman, Heidelberger, and Shahabuddin (1998a) to pricing of American options with a stochastic mesh. Third, we sketch generalizations of the mesh method and we discuss links with other techniques for valuing American options. Our empirical results show that the biasreduced point estimates are much more accurate than the standard meshmethod point estimates. Importance sampling is found to increase accuracy for a smooth optionpayoff functions, while variance increases are possible for nonsmooth payoffs.  
Return to
Wintersim 
STRATIFICATION ISSUES IN
ESTIMATING VALUEATRISK 


Paul
Glasserman Columbia University New York, NY 10027 
Philip
Heidelberger IBM T.J. Watson Research Center Yorktown Heights, NY 10598 
Perwez
Shahabuddin Columbia University New York, NY 10027 
ABSTRACT  
This paper considers efficient estimation of valueatrisk, which is an important problem in risk management. The valueatrisk is an extreme quantile of the distribution of the loss in portfolio value during a holding period. An effective importance sampling technique is described for this problem. The importance sampling can be further improved by combining it with stratified sampling. In this setting, an effective stratification variable is the likelihood ratio itself. The paper examines issues associated with the allocation of samples to the strata, and compares the effectiveness of the combination of importance sampling and stratified sampling to that of stratified sampling alone.  
Return to
Wintersim 
AN ASYMPTOTIC ALLOCATION
FOR SIMULTANEOUS SIMULATION EXPERIMENTS 


HsiaoChang
Chen SynQuest Inc. Clifton Park, NY 12065, U.S.A. 
ChunHung
Chen Jianwu Lin Dept. of Systems Engineering University of Pennsylvania Philadelphia, PA 19104, U.S.A.  
ABSTRACT  
In this paper, we consider the allocation of a fixed total number of simulation replications among competing design alternatives in order to (i) identify the best simulated design, (ii) intelligently determine the best simulation run lengths for all simulation experiments, and (iii) significantly reduce the total computation cost. An asymptotically optimal allocation rule for maximizing a lower bound of the probability of correct selection is presented. Moreover, we illustrate the efficiency of our method with a series of generic numerical experiments. The simulation cost is significantly reduced with our sequential approach.  
Return to
Wintersim 
Can the Regenerative
Method be Applied to DiscreteEvent Simulation



Shane G. Henderson
Department of Engineering Science University of Auckland Private Bag 92019 Auckland, NEW ZEALAND 
Peter W.
Glynn Department of EngineeringEconomic Systems and Operations Research Stanford University Stanford CA 943094023, U.S.A.  
ABSTRACT  
The regenerative method enjoys
asymptotic properties that make it a highly desirable approach for
steadystate simulation output analysis. It has been shown that virtually
all discreteevent simulations are regenerative. However, the method is
not in widespread use, perhaps primarily because of a difficulty in
identifying regeneration times. Our goal in this paper is to highlight the essence of the difficulty in identifying regeneration times in discreteevent simulations. We focus on a very simple example of a discreteevent simulation, and explore its regenerative properties. We show that for our example, it is possible to explicitly determine regeneration times. The ideas that are used to establish this fact might prove useful in identifying regeneration times in more general discreteevent system simulations.  
Return to
Wintersim 
FAST SIMULATION OF
BROADBAND TELECOMMUNICATIONS NETWORKS CARRYING LONGRANGE DEPENDENT BURSTY
TRAFFIC 


José R.
Gallardo Dimitrios Makrakis Advanced Communications Engineering Centre Department of Electrical and Computer Engineering The University of Western Ontario London, Ontario N6A 5B9, CANADA 
Luis
OrozcoBarbosa School of Information Technology and Engineering Department of Electrical Engineering University of Ottawa Ottawa, Ontario K1N 6N5, CANADA  
ABSTRACT  
A technique for the fast simulation of broadband communications systems is proposed, which is based on regenerative Importance Sampling techniques. Our algorithm is applicable to estimate the probability of rare events when modeling the offered traffic using Fractional Stable Noise (FSN) processes (including Fractional Brownian Noise as a particular case), which have been recently proved to be able to capture both the longrange dependence and the burstiness of today's aggregate network traffic. An exact description of FSN processes is given, as well as an approximation that allows for the application of Importance Sampling techniques. The results obtained for a simple example are also included.  
Return to
Wintersim 
SELECTION PROCEDURES
WITH STANDARDIZED TIMES SERIES VARIANCE ESTIMATORS



David Goldsman
William S. Marshall School of Industrial and Systems Engineering Georgia Institute of Technology Atlanta, GA 30332, U.S.A. 

ABSTRACT  
This article studies a modification of Rinott's twostage procedure for selecting the normal population with the largest (or smallest) mean. The modification, which is appropriate for use in the simulation environment, uses in the procedure's first stage different variance estimators than the usual batch means (BM) variance estimator. In particular, we will use variance estimators arising from the method of standardized time series (STS). On the plus side, certain STS estimators have more degrees of freedom than that of the BM estimator. On the other hand, STS variance estimators tend to require larger sample sizes than the BM estimator in order to converge to their assumed distributions. These considerations result in tradeoffs involving the procedure's achieved probability of correct selection as well as the procedure's expected sample size.  
Return to
Wintersim 
DESIGNING SIMULTANEOUS
SIMULATION EXPERIMENTS 


Paul
Hyden School of Operations Research and Industrial Engineering Cornell University 206 Rhodes Hall Ithaca, NY 14853, U.S.A 
Lee
Schruben Department of Industrial Engineering and Operations Research University of California at Berkeley 4135 Etcheverry Hall Berkeley, CA 947201777, U.S.A  
ABSTRACT  
Simulation experiments are often designed assuming that a fixed, and known, computing budget is to be allocated sequentially among different alternatives. However, in actual simulation experiments, there may be budget uncertainty or at least flexibility  for example, when there is a soft deadline for obtaining the study results. In such situations, it may be beneficial to allocate resources simultaneously in dynamically changing proportions. In this paper, we will examine optimal resource allocation paths. These paths climb the contour curves of the probability of selecting the best of several alternatives in a manner that insures that the highest probability of correct selection P(CS) is obtained when the study is halted. To gain insight into the complexity of optimal resource allocation paths, simple models exhibiting serial correlation, cross correlation, and trends are studied.  
Return to
Wintersim 
SIMULTANEOUS SIMULATION
EXPERIMENTS AND NESTED PARTITION FOR DISCRETE RESOURCE ALLOCATION IN
SUPPLY CHAIN MANAGEMENT 


Leyuan
Shi Dept. of Industrial Engineering University of WisconsinMadison Madison, WI 53706 
ChunHung
Chen Dept. of Systems Engineering University of Pennsylvania Philadelphia, PA 191046315 
Enver
Yûcesan INSEAD Technology Management Area Fontainebleau, France 
ABSTRACT  
Discrete resource allocation is a common problem in supply chain management. However, stochastic discrete resource allocation problems are difficult to solve. In this paper, we propose a new algorithm for solving such difficult problems. The algorithm integrates the nested partitions method with an optimal computing budget allocation method. The resulting hybrid algorithm retains the global perspective of the nested partitions method and the efficient simultaneous simulation experiments of the optimal computing budget allocation. Numerical results demonstrate that the hybrid algorithm can be effectively used for a largescale discrete resource allocation problem.  
Return to
Wintersim 
Rare Event Simulation of
Delay in Packet Switching Networks Using DPRBased Splitting



Zsolt
Haraszti Ericsson Radio Systems AB S12625 Stockholm, Sweden 
J. Keith
Townsend Center for Advanced Computing and Communication Department of Electrical & Computer Engineering North Carolina State University Raleigh, NC, 276957914 U.S.A.  
ABSTRACT  
Rare event simulation using splitting has been shown to provide significant speedup for large classes of problems, especially when queue length distribution is of primary interest. However, choice of the control parameters is much less straightforward in cases where splitting is applied to systems in which the target event is delay, rather than packet loss. In this paper we propose a control strategy for splitting that allows computationally efficient analysis of very low delay threshold probabilities which typically occur in communication networks. A different technique is required for delay because unlike the cell or packet loss case, the target event (delay) and the prerequisite condition that leads to a rare delay event (a full buffer) do not coincide temporally. We demonstrate the technique by using it to measure delay probabilities in three examples: a simple ATM multiplexer, a queueing system with multiple traffic classes, and a tandem queueing network with tagged and background traffic.  
Return to
Wintersim 
Efficient Simulation of
a Tandem Jackson Network 


Dirk P.
Kroese Teletraffic Research Center University of Adelaide South Australia 5005 
Victor F.
Nicola Department of Electrical Engineering University of Twente Enschede, The Netherlands  
ABSTRACT  
In this paper we consider a twonode
tandem Jackson network. Starting from a given state, we are interested in
estimating the probability that the content of the second buffer exceeds
some high level L before it becomes empty. The theory of Markov
additive processes is used to determine the asymptotic decay rate of this
probability, for large L. Moreover, the optimal exponential change
of measure to be used in importance sampling is derived and used for
efficient estimation of the rare event probability of interest.
Unlike changes of measures proposed and studied in recent literature, the one derived here is a function of the content of the first buffer, and yields asymptotically efficient simulation for any set of arrival and service rates. The relative error is bounded independent of the level L, except when the first server is the bottleneck and its buffer is infinite, in which case the relative error is bounded linearly in L.  
Return to
Wintersim 
SIMULATING HEAVY TAILED
PROCESSES USING DELAYED HAZARD RATE TWISTING



Sandeep
Juneja Indian Insititute of Technology Delhi 110016, India. 
Perwez
Shahabuddin Columbia University New York, NY 10027, USA. 
Anurag
Chandra Massachusetts Institute of Technology Cambridge, MA 02139, USA. 
ABSTRACT  
Consider the problem of estimating the small probability that the maximum of a random walk exceeds a large threshold, when the process has a negative drift and the underlying random variables may have heavy tailed distributions. We consider one class of such problems that has applications in estimating the ruin probability associated with insurance claim processes with subexponentially distributed claim sizes, and in estimating the probability of large delays in single server queues with subexponentially distributed service times. Significant work has been done on analogous problems for the light tailed case (when the moment generating function exists in a neighborhood around zero, so that the tail decreases at an exponential rate or faster) involving importance sampling methods that use exponential twisting. However, for the subexponential case, moment generating functions do not exist in the pertinent regions making exponential twisting infeasible. In this paper we introduce importance sampling techniques where the new probability measure is using the change of measure obtained by twisting the hazard rate of the original distribution. For subexponential distributions this amounts to twisting at a subexponential rate. We also introduce the technique of "delaying" the change of measure and show that the combination of the two techniques produces asymptotically optimal estimates of the small probabilities mentioned above for a large class of subexponential distributions.  
Return to
Wintersim 
SIMULATIONBASED
ESTIMATION OF QUANTILES 


E. Jack Chen W. David Kelton Department of Quantitative Analysis and Operations Management University of Cincinnati Cincinnati, Ohio 45221, U.S.A. 

ABSTRACT  
This paper discusses implementation of a sequential quantileestimation algorithm for highly correlated steadystate simulation output. Our primary focus is on issues related to computational and storage requirements of order statistics. The algorithm can compute exact sample quantiles and process sample sizes up to several billion without storing and sorting the whole sequence. The algorithm dynamically increases the sample size so that the quantile estimated satisfies a prespecified precision requirement.  
Return to
Wintersim 
On the Implementation of
Smoothed Perturbation Analysis Estimator for a SingleServer Queue with
Multiple Vacations 


Takayuki
Takagi Department of Systems Science Kyoto University Kyoto 6068501, JAPAN 
Naoto
Miyoshi Department of Mathematical and Computing Sciences Tokyo Institute of Technology Tokyo 1528552, JAPAN  
ABSTRACT  
We consider to apply the perturbation analysis~(PA) to a singleserver queue with multiple server vacations. A major difficulty in the implementation of PA estimator for such queueing systems is that the introduced perturbations are propagated and accumulated continuously without any resetting. This fact may lead to the divergence of PA estimates even if the limiting distribution exists. We show that it is possible to construct a sequence of points on the observed sample path such that the perturbations are accumulated only between the two adjacent points. The key idea lies in constructing a perturbed path which is not on the same sample as the observed nominal path but is identical in probability law.  
Return to
Wintersim 
IMPROVED BATCHING FOR
CONFIDENCE INTERVAL CONSTRUCTION IN STEADYSTATE SIMULATION



Natalie M. Steiger
Department of Information Systems and Operations Management University of North Carolina at Greensboro Greensboro, NC 274026165, U.S.A. 
James R. Wilson
Department of Industrial Engineering North Carolina State University 2401 Stinson Drive Raleigh, NC 276957906, U.S.A.  
ABSTRACT  
We describe an improved batchmeans procedure for building a confidence interval on a steadystate expected simulation response that is centered on the sample mean of a portion of the corresponding simulationgenerated time series and satisfies a userspecified absolute or relative precision requirement. The theory supporting the new algorithm merely requires the output process to be weakly dependent (phimix\ing) so that for a sufficiently large batch size, the batch means are approximately multivariate normal but not necessarily uncorrelated. A variant of the method of nonoverlapping batch means (NOBM), the Automated Simulation Analysis Procedure (ASAP) operates as follows: the batch size is progressively increased until either (a) the batch means pass the von Neumann test for independence, and then ASAP delivers a classical NOBM confidence interval; or (b) the batch means pass the ShapiroWilk test for multivariate normality, and then ASAP delivers a corrected confidence interval. The latter correction is based on an inverted CornishFisher expansion for the classical NOBM ratio, where the terms of the expansion are estimated via an autoregressivemoving average time series model of the batch means. An experimental performance evaluation demonstrates the advantages of ASAP versus other widely used batchmeans procedures.  
Return to
Wintersim 
SENSITIVITY OF OUTPUT
PERFORMANCE MEASURES TO INPUT DISTRIBUTION SHAPE IN MODELING QUEUES  3:
REAL DATA SCENARIO 


Donald
Gross Department of Systems Engineering and Operations Research George Mason University Fairfax, VA 22030, U.S.A 

ABSTRACT  
This paper followson papers presented at the two previous WSC conferences on sensitivity of output measures to input distribution selection in queueing modeling. Here, a real situation is studied, where data on input distributions are utilized and distributions selected by two fitting packages, Arena Input Analyzer and ExpertFit. Empirical distributions made from histograms of the raw data itself , as well as the first two choices from Arena and ExpertFit are compared for this small bank queueing network model, showing that an output measure such as mean wait in queue is quite sensitive to input distribution choice.  
Return to
Wintersim 
SIMULATING A
NONSTATIONARY POISSON PROCESS USING BIVARIATE THINNING: THE CASE OF
"TYPICAL WEEKDAY" ARRIVALS AT A CONSUMER ELECTRONICS STORE



K. Preston White,
Jr. Department of Systems Engineering University of Virginia Charlottesville, VA 229032442, U.S.A. 

ABSTRACT  
We present a case study in which thinning is applied to simulate timevarying arrivals at a consumer electronics store. The underlying simulation was developed to support an analysis of new staffing schedules for retail sales associates, given proposed changes in store layout and operating procedures. A principal challenge was developing a modeling approach for customer arrivals, where it was understood that the arrival rate varied by timeofday and by dayoftheweek, as well as seasonally. An analysis of arrival data supported a conjectured "typical weekday" as one basic arrival model. For this model, arrivals were assumed to be nonstationary Poisson, with a piecewiselinear arrival rate independently modulated by hour and by day. Arrival data were filtered and independent hourly and daily thinning factors computed. In the simulation, potential arrivals were generated with a mean equal to the minimum average interarrival rate, determined from the average arrival count for the hour/day time block with unit thinning factors. Candidate arrivals were then thinned using a bivariate acceptance probability equal to the product of the corresponding hourly and daily thinning factors.  
Return to
Wintersim 
Sensitivity Analysis of
Simulation Output To Parameters of Nonhomogeneous Poisson Processes



Michael E. Kuhl
Department of Industrial and Manufacturing Systems Engineering Louisiana State University Baton Rouge, LA 708036409, U.S.A. 
Sun Ewe
Lim Department of Industrial and Manufacturing Systems Engineering Louisiana State University Baton Rouge, LA 708036409, U.S.A.  
ABSTRACT  
Nonhomogeneous Poisson processes (NHPPs) are frequently used in stochastic simulations to model nonstationary point processes. These NHPP models are often constructed by estimating the parameters of the rate function from one or more observed realizations of the process. This paper focuses on the degree of accuracy to which the rate function parameters of the NHPP need to be estimated such that the simulation output performance measures are not significantly different from performance measures that would be obtained for the underlying (true) process.  
Return to
Wintersim 
OPTIMIZATION OVER
DISCRETE SETS VIA SPSA 


László
Gerencsér Computer and Automation Institute Hungarian Academy of Sciences Kende 1317, Budapest, 1111, HUNGARY 
Stacy D.
Hill Applied Physics Laboratory John Hopkins University Laurel, MD 207236099, U.S.A. 
Zsuzsanna Váagó
Computer and Automation Institute Hungarian Academy of Sciences Kende 1317 Budapest, 1111, HUNGARY 
ABSTRACT  
A fixed gain version of the SPSA (simultaneous perturbation stochastic approximation) method for function minimization is developed and the error process is characterized. The new procedure is applicable to optimization problems over, the grid of points in with integer components. Simulation results and a closely related application, a resource allocation problem, is shortly described.  
Return to
Wintersim 
Solving Stochastic
Optimization Problems with Stochastic Constraints: An Application in
Network Design 


Gül
Gürkan CentER for Economic Research Tilburg University 5000 LE Tilburg The Netherlands 
A.
Yonca GE Corporate R & D One Research Circle Niskayuna, NY 12309 USA 
Stephen M.
Robinson Department of Industrial Engineering University of WisconsinMadison 1513 University Avenue Madison, WI 53706 USA 
ABSTRACT  
Recently samplepath methods have been successfully used in solving challenging simulation optimization and stochastic equilibrium problems. In this paper we deal with a variant of these methods to solve stochastic optimization problems with stochastic constraints. Using optimality conditions, we convert the problem to a stochastic variational inequality. We outline a set of sufficient conditions for the almostsure convergence of the method. We also illustrate an application by using the method to solve a network design problem. We find optimal arc capacities for a stochastic network (in which the demand and supply at each node is random) that minimize the sum of the capacity allocation cost and a measure of the expected shortfall in capacity.  
Return to
Wintersim 
ITERATIVE
RANKINGANDSELECTION FOR LARGESCALE OPTIMIZATION



Sigurdur
Ólafsson Department of Industrial and Manufacturing Systems Engineering Iowa State University 205 Engineering Annex, Ames, IA 50010 

ABSTRACT  
We develop a new algorithm for simulationbased optimization where the number of alternatives is finite but very large. Our approach draws on recent work in adaptive random search and from rankingandselection. In particular, it combines the nested partitions method for global optimization and Rinott's twostage rankingandselection procedure. We prove asymptotic convergence of the new algorithm under fairly mild conditions.  
Return to
Wintersim 
STATISTICAL METHODS FOR
SENSITIVITY AND PERFORMANCE ANALYSIS IN COMPUTER EXPERIMENTS



Leslie M.
Moore Statistical Sciences Group, TSA1, MS F600 Los Alamos National Laboratory Los Alamos, NM 875450600, USA 
Bonnie K.
Ray Dept. of Mathematical Sciences New Jersey Institute of Technology Newark, NJ 07102, USA  
ABSTRACT  
We describe statistical methods for sensitivity and performance analysis of complex computer simulation experiments. Graphical methods, such as trellis plots, are suggested for exploratory analysis of individual or aggregate performance metrics conditional on different experiment inputs. More formal statistical methods, such as analysis of variancebased methods and regression tree analysis, are used to determine variables having substantive influence on the experimental results and to investigate the structure of the underlying relationship between inputs and outputs. The methods are discussed in relation to a supply chain model of the textile manufacturing process having many possible input and output variables of interest and for a computer model used to describe the flow of material in an ecosystem.  
Return to
Wintersim 
A SURVEY OF RANKING,
SELECTION, AND MULTIPLE COMPARISON PROCEDURES FOR DISCRETEEVENT
SIMULATION 


James R.
Swisher Lotus Biochemical Corporation 7335 Lee Highway Radford, Virginia 24141 U.S.A. 
Sheldon H.
Jacobson Department of Mechanical and Industrial Engineering University of Illinois at UrbanaChampaign Urbana, Illinois 61801 U.S.A.  
ABSTRACT  
Discreteevent simulation models are often constructed so that an analyst may compare two or more competing design alternatives. This paper presents a survey of the literature for two widelyused statistical methods for selecting the best design from among a finite set of k alternatives: ranking and selection (R&S) and multiple comparison procedures (MCPs). A comprehensive survey of each topic is presented along with a summary of recent unified R&SMCP approaches. In addition, an example of the application of Nelson and Matejcik's (1995) combined R&SMCP procedure is given.  
Return to
Wintersim 
The Main Issues in
Nonlinear Simulation Metamodel Estimation



Isabel Reis dos
Santos Departamento de Matemática Instituto Superior Técnico Av. Rovisco Pais, 1049 Lisboa, PORTUGAL 
Acácio M. O. Porta
Nova Seccào Autónoma de Economia e Gestào Instituto Superior Técnico Av. Rovisco Pais, 1049 Lisboa, PORTUGAL  
ABSTRACT  
In this paper, we investigate and discuss some of the main issues concerning the estimation of nonlinear simulation metamodels. We propose a methodology for identifying a tentative functional relationship, estimating the metamodel coefficients and validating the simulation metamodel. This approach is illustrated with a simple queueing system. Finally, we draw some conclusions and identify topics for further work in this area.  
Return to
Wintersim 