ISYE 6644 OH EXAM 2 PRACTICE QUESTIONS – SPRING 2025
COMPLETE STUDY GUIDE WITH VERIFIED QUESTIONS,
ANSWERS & RATIONALES GEORGIA INSTITUTE OF
TECHNOLOGY REAL 150 QUESTIONS EDITION: 2026 INSTANT
DOWNLOAD PDF
CORE DOMAINS COVERED IN THIS EXAM:
• Output Analysis & Statistical Estimation: Finite-horizon vs. steady-
state simulations, independent replications, sample mean variance, and
confidence intervals.
• Initialization Bias & Data Truncation: Methods for dealing with
initialization bias, warm-up periods, and run length determination.
• Variance Reduction Techniques: Common random numbers,
antithetic variates, control variates, and importance sampling.
• Random Number Generation: Linear congruential generators,
properties, and testing for randomness.
• Random Variate Generation: Inverse transform method, acceptance-
rejection method, and convolution.
• Input Modeling: Fitting distributions, parameter estimation, goodness-
of-fit tests, and empirical distributions.
• Simulation Optimization: Response surface methodology, steepest
descent, and metaheuristics.
• Comparing Systems: Statistical comparison of alternative system
designs, paired-t confidence intervals, and ranking/selection
procedures.
SECTION 1: OUTPUT ANALYSIS AND STATISTICAL ESTIMATION
(Questions 1-30)
,Question 1: We often distinguish between two general types of
simulations with regard to output analysis. What are they called?
A. Discrete and continuous
B. Deterministic and stochastic
C. Finite-horizon and steady-state
D. Terminating and non-terminating
Rationale: Finite-horizon (terminating) simulations have a natural stopping
time (e.g., 8:00 a.m. to 5:00 p.m.), while steady-state (non-terminating)
simulations run indefinitely to analyze long-run behavior.
Question 2: Which scenario is best suited for a finite-horizon analysis?
A. Simulate an assembly line working 24/7
B. Simulate bank operations from 8:00 a.m. to 5:00 p.m.
C. Simulate a manufacturing plant with continuous production
D. Simulate a call center that never closes
Rationale: Bank operations from opening to closing have a natural
termination time, making it a finite-horizon simulation.
Question 3: Which scenario is best suited for a steady-state analysis?
A. Simulate a bank from 9:00 a.m. to 5:00 p.m.
B. Simulate a one-day sale event
C. Simulate an assembly line working 24/7
D. Simulate a single customer visit
Rationale: An assembly line running continuously (24/7) is analyzed for
long-run steady-state behavior.
Question 4: TRUE or FALSE? The main method of attack for terminating
simulations is via independent replications.
A. True
B. False
,Rationale: Independent replications help mitigate issues with simulation
outputs not being i.i.d., providing valid confidence intervals.
Question 5: TRUE or FALSE? Suppose that X₁, X₂, ..., Xₙ are consecutive
waiting times, and we define the sample mean X̄ = ΣXᵢ/n. Then Var(X̄ ) =
Var(Xᵢ)/n.
A. True
B. False
Rationale: Very FALSE! Correlation between observations messes up the
variance of the sample mean. This is one of the main reasons why output
analysis is difficult.
Question 6: How can we deal with initialization bias if we want to do a
steady-state analysis?
A. Make an extremely long run to overwhelm it
B. Truncate (delete) some of the initial data
C. Use independent replications
D. Both A and B
Rationale: Initialization bias can be mitigated by running a long simulation to
overwhelm the bias or by truncating initial observations to remove the
transient phase.
Question 7: TRUE or FALSE? In a steady-state simulation, the output data
are typically independent and identically distributed (i.i.d.).
A. True
B. False
Rationale: Output data from steady-state simulations are typically correlated
(autocorrelated), not independent.
Question 8: The "batch means" method is used to:
, A. Generate random numbers
B. Estimate the variance of the sample mean in steady-state simulations
C. Reduce simulation runtime
D. Generate random variates
Rationale: The batch means method divides a long simulation run into
batches and treats the batch means as approximately independent to estimate
the variance of the overall sample mean.
Question 9: TRUE or FALSE? In a terminating simulation, confidence
intervals are easier to construct because we can assume independence
across replications.
A. True
B. False
Rationale: In terminating simulations, replications start at the same initial
state, and the outputs across replications are independent, making confidence
interval construction straightforward.
Question 10: Which of the following is a problem with output analysis in
steady-state simulations?
A. Non-stationary output
B. Autocorrelation
C. Initialization bias
D. All of the above
Rationale: Steady-state simulations often have correlated outputs, non-
stationary transient behavior, and initialization bias.
Question 11: A "warm-up period" is:
A. The time before the simulation starts
B. The initial period of a simulation that is discarded to eliminate
initialization bias
COMPLETE STUDY GUIDE WITH VERIFIED QUESTIONS,
ANSWERS & RATIONALES GEORGIA INSTITUTE OF
TECHNOLOGY REAL 150 QUESTIONS EDITION: 2026 INSTANT
DOWNLOAD PDF
CORE DOMAINS COVERED IN THIS EXAM:
• Output Analysis & Statistical Estimation: Finite-horizon vs. steady-
state simulations, independent replications, sample mean variance, and
confidence intervals.
• Initialization Bias & Data Truncation: Methods for dealing with
initialization bias, warm-up periods, and run length determination.
• Variance Reduction Techniques: Common random numbers,
antithetic variates, control variates, and importance sampling.
• Random Number Generation: Linear congruential generators,
properties, and testing for randomness.
• Random Variate Generation: Inverse transform method, acceptance-
rejection method, and convolution.
• Input Modeling: Fitting distributions, parameter estimation, goodness-
of-fit tests, and empirical distributions.
• Simulation Optimization: Response surface methodology, steepest
descent, and metaheuristics.
• Comparing Systems: Statistical comparison of alternative system
designs, paired-t confidence intervals, and ranking/selection
procedures.
SECTION 1: OUTPUT ANALYSIS AND STATISTICAL ESTIMATION
(Questions 1-30)
,Question 1: We often distinguish between two general types of
simulations with regard to output analysis. What are they called?
A. Discrete and continuous
B. Deterministic and stochastic
C. Finite-horizon and steady-state
D. Terminating and non-terminating
Rationale: Finite-horizon (terminating) simulations have a natural stopping
time (e.g., 8:00 a.m. to 5:00 p.m.), while steady-state (non-terminating)
simulations run indefinitely to analyze long-run behavior.
Question 2: Which scenario is best suited for a finite-horizon analysis?
A. Simulate an assembly line working 24/7
B. Simulate bank operations from 8:00 a.m. to 5:00 p.m.
C. Simulate a manufacturing plant with continuous production
D. Simulate a call center that never closes
Rationale: Bank operations from opening to closing have a natural
termination time, making it a finite-horizon simulation.
Question 3: Which scenario is best suited for a steady-state analysis?
A. Simulate a bank from 9:00 a.m. to 5:00 p.m.
B. Simulate a one-day sale event
C. Simulate an assembly line working 24/7
D. Simulate a single customer visit
Rationale: An assembly line running continuously (24/7) is analyzed for
long-run steady-state behavior.
Question 4: TRUE or FALSE? The main method of attack for terminating
simulations is via independent replications.
A. True
B. False
,Rationale: Independent replications help mitigate issues with simulation
outputs not being i.i.d., providing valid confidence intervals.
Question 5: TRUE or FALSE? Suppose that X₁, X₂, ..., Xₙ are consecutive
waiting times, and we define the sample mean X̄ = ΣXᵢ/n. Then Var(X̄ ) =
Var(Xᵢ)/n.
A. True
B. False
Rationale: Very FALSE! Correlation between observations messes up the
variance of the sample mean. This is one of the main reasons why output
analysis is difficult.
Question 6: How can we deal with initialization bias if we want to do a
steady-state analysis?
A. Make an extremely long run to overwhelm it
B. Truncate (delete) some of the initial data
C. Use independent replications
D. Both A and B
Rationale: Initialization bias can be mitigated by running a long simulation to
overwhelm the bias or by truncating initial observations to remove the
transient phase.
Question 7: TRUE or FALSE? In a steady-state simulation, the output data
are typically independent and identically distributed (i.i.d.).
A. True
B. False
Rationale: Output data from steady-state simulations are typically correlated
(autocorrelated), not independent.
Question 8: The "batch means" method is used to:
, A. Generate random numbers
B. Estimate the variance of the sample mean in steady-state simulations
C. Reduce simulation runtime
D. Generate random variates
Rationale: The batch means method divides a long simulation run into
batches and treats the batch means as approximately independent to estimate
the variance of the overall sample mean.
Question 9: TRUE or FALSE? In a terminating simulation, confidence
intervals are easier to construct because we can assume independence
across replications.
A. True
B. False
Rationale: In terminating simulations, replications start at the same initial
state, and the outputs across replications are independent, making confidence
interval construction straightforward.
Question 10: Which of the following is a problem with output analysis in
steady-state simulations?
A. Non-stationary output
B. Autocorrelation
C. Initialization bias
D. All of the above
Rationale: Steady-state simulations often have correlated outputs, non-
stationary transient behavior, and initialization bias.
Question 11: A "warm-up period" is:
A. The time before the simulation starts
B. The initial period of a simulation that is discarded to eliminate
initialization bias
