BNAL 476 EXAM 2 QUESTIONS WITH VERIFIED SOLUTIONS LATEST UPDATE 2026
Central Limit Theorem - Answers As sample size gets larger the distribution of the sample means
approaches the normal distribution
Point Estimates - Answers Single number and doesn't indicate uncertainty or variablitity, associated
with it
Confidence Interval - Answers A range, upper and lower bounds, plus the CI %
Null hypothesis for simulation - Answers H0: what is to be tested
Alternative hypothesis for simulation - Answers H1: denial of H0
low p-value for hypothesis testing (like <.01) - Answers Reject H0
high p-value for hypothesis testing (like >.2) - Answers Accept H0
Type 1 error - Answers Reject H0 but H0 is really true
Type 2 error - Answers Don not reject H0 but H0 is really false
Use Hypothesis Test for simulation Input side - Answers Fit a probability distribution to the real-world
data
Use Hypothesis Test for simulation Output side - Answers Selection of a "best" model scenario
To calculate number of scenarios - Answers factors multiplied by value number ex: 3 factors 5 value
(5x5x5)
Terminating systems - Answers Definite end-time. Idle and Empty: day starts maybe no customer
shows up right when they open
Steady State systems - Answers Warm-up period and batched means (long single run is broken up
into replications)
Determine number of replications - Answers n = (z,a/2 s/h)^2 and h= z,a/2 s/sqrt(2). Need to find s
(variance)
Theoretical distribution - Answers Normal, Expo, and Trianglar, etc.
Fitting decisions - Answers Distributions form
Distributions parameters
Distributions form - Answers exponential, gamma, empiraical, etc.
Distributions parameters - Answers mean, std dev, lambda, etc
Fitting distributions for data hypothesis - Answers H0: distribution is not a good fit
H1: distribution is a good fit
Fitting distributions methods - Answers Mean square error, chi-square goodness of fit, Kolmogorov-
Smirnov
Pitfalls of normal distribution - Answers Has infinite tails in both directions
Input Analyzer is used - Answers find best distribution fit for data
How to pick best distribution through Input Analyzer - Answers high p-values and >.15 KS test p-value
Verification - Answers Is "Code" = Model. Does computer model match conceptual model
Validation - Answers Is Model=System does the conceptual model accurately depict real system
Attempt for verification - Answers eliminate error messages, restrict # of entities, replace some
distributions w/ constant the compare, "stress" model under extreme conditions
Validation techniques - Answers Conceptual model validation and plots and animation
Conceptual model validation - Answers Structured walk-through w/ project managers, analysts, and
domain experts
Validation techniques if there is an existing system - Answers Compare performance measures of
model vs. historic data, this is called retrodiction, backtesting, or "results validation"
Validation techniques if no existing system - Answers Analysts & domain experts review for
reasonableness. "Force validity"
Issue of stationarity (validation issue) - Answers Non-stationary and auto-correlation are common in
both real systems and sim models. Model only approximates the real system, so H0 is already false
Simulation Optimization - Answers tries to combine the benefits of simulation modeling with
simulation optimization
Simulation Optimization - Answers Substitutes simulation model for objective function equation
Simulation Optimization - Answers Handles non-linearity and stochastic processes
Simulation Optimization - Answers Uses inteligent search methods to change model factors run
simulation, evaluate results, make new changes, rerun. Use OptQuest in Arena.
Null hypothesis for Welch CI - Answers H0: mean1-mean2=0 (no difference)
Alternative hypothesis for Welch CI - Answers H1: mean1-mean2 doesn't =0 (difference)
Central Limit Theorem - Answers As sample size gets larger the distribution of the sample means
approaches the normal distribution
Point Estimates - Answers Single number and doesn't indicate uncertainty or variablitity, associated
with it
Confidence Interval - Answers A range, upper and lower bounds, plus the CI %
Null hypothesis for simulation - Answers H0: what is to be tested
Alternative hypothesis for simulation - Answers H1: denial of H0
low p-value for hypothesis testing (like <.01) - Answers Reject H0
high p-value for hypothesis testing (like >.2) - Answers Accept H0
Type 1 error - Answers Reject H0 but H0 is really true
Type 2 error - Answers Don not reject H0 but H0 is really false
Use Hypothesis Test for simulation Input side - Answers Fit a probability distribution to the real-world
data
Use Hypothesis Test for simulation Output side - Answers Selection of a "best" model scenario
To calculate number of scenarios - Answers factors multiplied by value number ex: 3 factors 5 value
(5x5x5)
Terminating systems - Answers Definite end-time. Idle and Empty: day starts maybe no customer
shows up right when they open
Steady State systems - Answers Warm-up period and batched means (long single run is broken up
into replications)
Determine number of replications - Answers n = (z,a/2 s/h)^2 and h= z,a/2 s/sqrt(2). Need to find s
(variance)
Theoretical distribution - Answers Normal, Expo, and Trianglar, etc.
Fitting decisions - Answers Distributions form
Distributions parameters
Distributions form - Answers exponential, gamma, empiraical, etc.
Distributions parameters - Answers mean, std dev, lambda, etc
Fitting distributions for data hypothesis - Answers H0: distribution is not a good fit
H1: distribution is a good fit
Fitting distributions methods - Answers Mean square error, chi-square goodness of fit, Kolmogorov-
Smirnov
Pitfalls of normal distribution - Answers Has infinite tails in both directions
Input Analyzer is used - Answers find best distribution fit for data
How to pick best distribution through Input Analyzer - Answers high p-values and >.15 KS test p-value
Verification - Answers Is "Code" = Model. Does computer model match conceptual model
Validation - Answers Is Model=System does the conceptual model accurately depict real system
Attempt for verification - Answers eliminate error messages, restrict # of entities, replace some
distributions w/ constant the compare, "stress" model under extreme conditions
Validation techniques - Answers Conceptual model validation and plots and animation
Conceptual model validation - Answers Structured walk-through w/ project managers, analysts, and
domain experts
Validation techniques if there is an existing system - Answers Compare performance measures of
model vs. historic data, this is called retrodiction, backtesting, or "results validation"
Validation techniques if no existing system - Answers Analysts & domain experts review for
reasonableness. "Force validity"
Issue of stationarity (validation issue) - Answers Non-stationary and auto-correlation are common in
both real systems and sim models. Model only approximates the real system, so H0 is already false
Simulation Optimization - Answers tries to combine the benefits of simulation modeling with
simulation optimization
Simulation Optimization - Answers Substitutes simulation model for objective function equation
Simulation Optimization - Answers Handles non-linearity and stochastic processes
Simulation Optimization - Answers Uses inteligent search methods to change model factors run
simulation, evaluate results, make new changes, rerun. Use OptQuest in Arena.
Null hypothesis for Welch CI - Answers H0: mean1-mean2=0 (no difference)
Alternative hypothesis for Welch CI - Answers H1: mean1-mean2 doesn't =0 (difference)
