Simulation Analysis
QA 772 (22-QA-772-901)
Winter Quarter 2006, Mondays 6:00–8:40p.m., 216 Lindner
Web site for the class:  http://www.cba.uc.edu/faculty/keltonwd/qa772/

Syllabus

• Instructor

David Kelton, david.kelton@uc.edu, http://www.cba.uc.edu/faculty/keltonwd/, 556-6834, 525 Lindner
Office Hours: Before and after each class and pretty much all the time by e-mail

• Required Materials

• Law and Kelton, Simulation Modeling and Analysis, 3d. ed., McGraw-Hill, 2000, ISBN 0-07-059292-6
• Course notes (available at DuBois Bookstore ONLY ... obtain these before the first class on Jan. 9 and bring them to every class)

• Objectives

This is a second course in computer simulation. Study the underlying probabilistic and statistical aspects of computer simulation. Modeling and estimating input processes, random-number generators, variate and process generation, statistical analysis of simulation output, ranking and selection of simulation models, variance-reduction techniques, designing simulation experiments, gradient estimation, and optimization. By the end of the course, students should be able to start reading (and contributing to) the simulation research literature. The course will not cover simulation modeling (see the first prerequisite below).

• Required Prerequisites (all are firm)

• A first course in simulation modeling (with a general-purpose language or a simulation language).  This must have been an entire course devoted to simulation modeling; it is not sufficient to have had a brief introduction to simulation as part of an operations-research survey course.
• Probability and statistics (calculus-based).
• Computer and programming skills (FORTRAN, C, C++, etc.; spreadsheeting is not enough).

• Grading

• Mid-term Exam (Monday Feb. 13, 6:00–8:40p.m.) = 30%
• Final Exam (Mon. March 13, 6:00–8:40p.m.) = 40%
• Individual Project/ Presentation/Paper = 30% (more information below)

Both the Mid-term and Final Exams are open-book; no other materials will be permitted. Homework will be assigned but not collected; collaboration on the homework is fine, even encouraged.

• Computing

You may use any computer you like, either your own or in a university lab or office, provided that it has the software you'll need.  You may obtain the Arena 7.01.00 simulation software (academic version), from the instructor (just ask).

• Topics

• Introduction and overview of simulation analysis
• Modeling and estimating input processes (Chapter 6)
• Random-number generation (Chapter 7)
• Generation of random variates, vectors, and processes (Chapter 8)
• Statistical analysis of simulation output (Chapter 9)
• Comparison, ranking, and selection of simulation models (Chapter 10)
• Variance-reduction techniques (Chapter 11)
• Designing simulation experiments, gradient estimation, and optimization (Chapter 12)

Specific reading and homework assignments will be made in each class, and will depend on our progress.

• Individual Project Information

This will be a simulation application or methodological study of your own choosing. It should be a study with one of the following two approaches:

• Application of appropriate and effective simulation methodology in another field; e.g., manufacturing, health care, logistics, or telecommunications (many others are possible).
• A methodological study about simulation itself; e.g., comparing two different algorithms for ranking and selection on the basis of accuracy or efficiency.

This project is individual — it must be done by yourself without consultation with anyone but the instructor. The project must involve some amount of computer work. The "product" is a written paper and a presentation in class on the last class day.

Milestones

• Tuesday Feb. 6: Submit a one-"page" (via e-mail to the instructor) proposal of what you plan to do.
• Monday Feb. 20: Submit a one-"page" (via e-mail)  progress report.
• Monday March 6: Presentation in class.
• Wednesday March 15: Written report due by 6:00p.m. (via e-mail to the instructor).

• Schedule

• Links

• INFORMS Simulation Society
• Winter Simulation Conference
• Full WSC papers from 1997 - 2005
• EUROSIM, the Federation of European Simulation Societies
• Rockwell Software (formerly Systems Modeling Corporation), the vendor of Arena
• Simulation of the Buffon Needle Problem
• The original "bug" and more on Grace Murray Hopper (who coined the term)
• Don't even think about daring to click here if you are too young to have used punchcards