Deterministic Models in Operations Research
Fall 2004, Wednesday 7:20 Ė 10:00 p.m., Robinson A106
Dr. Kirk A.Yost
National Security Analysis Group, MITRE Corp.
phone: (703) 883-3133 (days)
office hours: to be announced, or by appointment
electronic mail: OR541GMU@aol.com
Wayne L. Winston, Operations Research, Applications and Algorithms, Third Edition, Duxbury Press, 1994.
This course will introduce deterministic Operations Research methods and applications, concentrating on linear, network, and integer programming methods. In addition, the course offers a brief overview of nonlinear optimization, and, time permitting, a discussion of recent developments in optimization techniques. Course content will concentrate on the applicability, assumptions, limitations, and solution methodologies of these methods, with the aim of teaching the art of formulating real world-problems using these types of models. While the principal goals are to provide the student with solid skills in formulation and solution analysis, certain math skills are necessary to fully understand these techniques. In particular, a working knowledge of linear algebra and multivariate calculus is required; this material will not be covered in this course.
Over the last decade, tremendous improvements in commercial solution packages and high-speed computing availability have made it possible for anyone with the appropriate knowledge to formulate and analyze large and difficult optimization problems that were unsolvable as recently as the early 90ís. To this end, we will require the use of the algebraic modeling language MPL to formulate and solve homework and project problems. A student version of the MPL model generator, along with the CPLEX linear and integer solver, can be downloaded from Maximal Software.
We will have two in-class exams: a midterm (20%) and a
cumulative final (35%). We will grade student homework (20%) to promote keeping
up with the material. A course project emphasizing application and presentation
(25%) will be due in the second half of the course.
Reading and Homework Assignments
Midterm: Monday, October 21
Final: Monday, December 20, 7:30-10:15 p.m.
All exams will be open book, open notes.
∑ While successful Operations Research involves a team approach, it is necessary that the analyst member of the team (eventually you, the student) be competent in the methods you are applying. Consequently, I expect you to do your own work on homework and the project. If I detect rampant collusion, I will decide on the quality of the work and distribute the points earned for a single attempt among the collaborators. Example: two people give me submissions that are obviously duplicates, and I decide itís a 90% effort. Each of the two students either get 45%, or they divide the 90% among themselves in some equitable manner.
∑ This syllabus contains tentative homework assignments; I reserve the right to change them depending on the pace of the course. Homework is due at the start of the class, and there is a 10% penalty per day for late submissions. In general, any homework assignment will be due after Iíve covered the relevant material in class. I also reserve the right to deduct points of illegibility and particularly for incomprehensibility, including grammar and spelling. You cannot be a successful analyst if you cannot communicate.
∑ With respect to communication, I will require that as part of your project, you prepare a PowerPoint presentation that describes your results, in a general format appropriate for a high-level decision-maker. Communication is critical in Operations Research, and so much good work fails to get implemented due to poor presentation that I am making this a firm requirement in this course. No decision-maker will accept your work if you cannot explain, in a comprehensible and economical way, the insights and implications of your analyses.
∑ I am entirely reasonable with respect to emergencies and other situations (such as employer demands) that are beyond your control. I will work with you provided you give me some advance notice.
∑ I expect you to show up for the lectures, and to be prepared to discuss the material. This means that you should read the assigned chapters beforehand and come armed with some questions. The size of the class (30+ students) will limit the amount of interaction per student, but I do not intend for this course to be a spectator sport.
For many of you, this will be your first course in Operations Research. As such, itís my job to ensure that you find it interesting, challenging, and even fun, so that youíll want to continue in the field. Iím a career OR analyst, with 22 years of experience in a wide range of areas. As a result, I have considerable knowledge of things that you generally donít find in textbooks, and I will pass that knowledge to you as best I can. I will stay close to the textbook that we are using; itís a good one, and is written at an appropriate level. I will supplement the text, however, and also give you advice the relative importance of various areas we cover. I urge you to take advantage of the lessons I have learned (sometimes painfully) over the past two decades.
Remember: a model is an abstraction of reality. As such, all models are wrong. Some, however, are useful. The objective of this course is to provide you with your initial training on an important class of OR models, and to give you the knowledge to first, recognize when they are useful, and second, apply them properly.
1. Linear Programming by George Dantzig
2. Top Ten Secrets to Success with Optimization by Gerald Brown
MPL Modeling System
Getting a computer account
IT&E Computer Labs (schedules, software, etc.)