CS 590M: Introduction to Simulation

Syllabus and Lecture Schedule

Prof. Peter J. Haas

Course Goals:

The emphasis will be on understanding the underlying principles and basic techniques of simulation modeling and analysis, so that students can apply simulation in a flexible and intelligent manner to real-world problems, and become educated consumers of simulation studies and simulation packages. Students will learn to appreciate the power and scope of simulation in applications drawn from a variety of domains.

Course Objectives:

Students will be able to

Use basic Monte Carlo methods to estimate probabilities, expected values, and more, that are hard to compute analytically or numerically
Formulate a mathematically precise stochastic simulation model from a description of an existing or proposed real-world system.
Use existing data or theoretical guidance to determine appropriate input distributions for the random components of a simulation model.
Determine appropriate algorithms for generating realizations of a simulation model.
Efficiently implement the algorithms on a computer in Python.
Statistically analyze the output of the simulation program to estimate system properties of interest, and to identify optimal systems designs and operating policies.
Demonstrate understanding of the mathematical principles underlying simulation methodology.
Critique someone else’s simulation results.

Readings

For some topics, there will be readings (and problems) taken from the course textbook, Simulation Modeling and Analysis by Averill Law. Other topics will be covered by the annotated slides that will be posted.

Exams

Midterm exam #1: Thursday, February 20, 7-9pm, ELAB 303
Midterm exam #2: Thursday, March 26, 7-9pm, ELAB 303
Final exam: Thursday, May 7, 10:30am-12:30pm, Goessmann 51

Lecture schedule (approximate):

Date	Topic	Notes
Week of 1/20	Introduction: the power of simulation, simulation challenges; basic Monte Carlo; basic point and interval estimation	HW 1
Week of 1/27	Probability models for discrete-event systems: simulating Markov chains, simple generation of discrete random variables
Week of 2/3	Probability models for discrete-event systems: simulating Markov, semi-Markov and generalized semi-Markov processes; variable time advance mechanism; inversion method for generating continuous random variables	HW 2
Week of 2/10	Input distributions: theoretical guidance; maximum-likelihood parameter estimation; Bayesian parameter estimation
Week of 2/17	Generation of non-uniform random numbers: acceptance-rejection, composition, convolution, alias method	HW 3; Midterm 1
Week of 2/24	Generation of uniform random numbers: congruential generators, period length and number theory; pitfalls; modern generators; quality testing
Week of 3/2	Data structures for event lists: linked lists, heaps, hybrid structures	HW 4
Week of 3/9	Output analysis: Estimating nonlinear functions of means, quantiles	HW 5
Week of 3/16	Spring break
Week of 3/23	No lectures, assignment on agent-based simulation	Midterm 2
Week of 3/30	Steady-state simulation: regenerative and batch-means methods
Week of 4/6	Efficiency-improvement techniques: common random numbers, antithetic variates, conditional Monte Carlo, control variates,	HW 6
Week of 4/13	Intro to experimental design and simulation-based optimization, gradient estimation, Robbins-Monro algorithm
Week of 4/23	Discrete simulation-based optimization
Week of 4/27	Review
7-May	Final Exam 10:30am-12:30pm, Location TBD

Last modified 10 March 2020