next up previous
Next: Three Cautionary Tales

Michael A. Trick
Visiting Associate Professor

Location: E40-121, x3-0534 Secretary: Kathy Sullivan 258-5583, E53-350

Email: trick+@cmu.edu

Course Home Page: http://mat.gsia.cmu.edu/mstc/

Office Hours: TBA or by appointment

Grader/T.A.: TBA

Description: Consultants use management science techniques in two ways: either as implementors or as consumers. The first happens when the consultant faces a problem amenable to an appropriate management science technique. For most consultants, this happens regularly enough to justify knowing some basic techniques. The second use of management science happens very often, and generally leads to consultations with specialists in such techniques. In this case, a consultant needs to know when such a specialist may be useful and to be able to evaluate the advice of the specialist.

The purpose of this course is to provide useful information, insights, and case histories to support both uses of management science. There are three aspects to meeting these goals.

The first is to show how management science is actually used in practice. Journals in management science show many, many examples of practical use of MS techniques. In fact, if the success of these projects is to be believed, every MS project saves either 40 million dollars, or 30 percent of the total project cost, whichever is greater. Although these are naturally exaggerations, it is clear that there are tremendous savings available in many areas by adopting a formal, mathematical, decision making process. Furthermore, from these projects are general rules for applying these techniques. Some are just common sense (``don't take your data too seriously'' and ``solve the problem you have, not the one you think you have'') while others provide an intuitive interpretation of deep technical results. These are important both to implementors (so they don't mess up) and to consumers (so they see who is messing up).

The second facet of this course is the teaching of new tools. At this point, you all have learned the most basic management science technique: linear programming. In practice, however, these represent only a small portion of the work done. I have identifies some areas where I would like to get across some technique-oriented skills: large scale linear programming, integer programming, heuristic solution methods, network programming, and multiple objective decision making. These tools have been chosen for their usefulness and for the ease in which a typical consultant can use them. The emphasis is on the use of tools which every consultant has (primarily spreadsheets).

The final aspect of this course is that it is somewhat ``hands on.'' There will be readings for class discussion, and there will be projects to allow you to work on problems using the tools you have learned.

Format and Class Notes: There is no appropriate textbook. You will be receiving copies of journal articles, supplemented by book chapters and so on. These should be read as assigned, for they form the basis for class discussion. Journal articles tend to be fairly difficult to read, so do not get upset if you do not understand everything before class. There will be also complete class notes available, which will be handed out as the class progresses.

I have arranged for a number of papers to be copied, but I will not necessarily cover all of the readings in class. If there is some area of particular interest to you, do not hesitate to talk to me. I may be able to fit in an appropriate piece for the readings.

For the exam, you will be responsible only for what is covered in class and for what I explicitly assign from the readings.

Grading: It is not possible to understand management science without doing management science. Therefore, the bulk of the work in this course will be in miniprojects. There will be seven miniprojects assigned throughout the term. Each student must do at least six of these projects (if you do all seven, then your six highest grades will be used). You may do these projects in groups of no more than four people, though the miniprojects may be done alone. Each project is worth 10 There will also be a final exam worth 40%.

Late Policy: Miniprojects are due in class on the days given below. Unexcused late work will be penalized 20% per day.

Outline: This is a tentative outline, which might change to adapt to the interests of the class.

September 5. Introductions. What is Management Science and why should consultants care? Project 7 assigned.

September 10. Review of linear programming models. The Diet Problem

September 12. Optimization on a spreadsheet.

September 17. Large scale linear optimization. Optimal Leases at GE Capital.

September 19. Cutting stock and other large linear programs. Cutting Photographic Color Paper Rolls. Project 1 assigned

September 24. Integer programming modeling. Optimal Snow Disposal

September 26. More integer programming modeling. Homart Development

October 1. Solving integer programs. A hopeless exercise? Optimal Check Clearing. Project 1 due. Project 2 assigned.

October 3. Heuristic solution techniques. Solution creation Helicopter Scheduling

October 8. Heuristic solution techniques. Solution improvement. Class Scheduling

October 10. Simulated Annealing, Tabu search. Tabu Search: A Tutorial. Project 2 due. Project 3 assigned.

October 15. NO CLASS (Columbus Day)

October 17. Genetic Algorithms. Genetic Algorithms as a tool for OR

October 22. Neural Networks. Neural Networks for the OR/MS Analyst

October 24. Evaluating Heuristics. Meals on Wheels Scheduling. Project 3 due. Project 4 assigned.

October 29. Relaxations. TBA

October 31. Strengthing Relaxations. TBA

November 5. NO CLASS (INFORMS)

November 7. Network Optimization. Municipal Street Sweeping Operations. Project 4 due. Project 5 assigned.

November 12. Special Cases of Network Optimization. Scheduling American League Umpires

November 14. Data Envelopment Analysis. Pupil Transportation in NC, Managing Bank Productivity

November 19. Stochastic Linear Programming Search for the SS Central America. Project 5 due. Project 6 assigned.

November 21. Multiple Objective Optimization. TBA

November 26. Analytical Heirarchy Process How to Make a Decision

November 28. NO CLASS (THANKSGIVING)

December 3. Putting it all together. Baseball Scheduling Project 6 due.

December 5. Putting it all together. TBA.

December 10. Putting it all together. Project 7 due.

FINAL EXAM

Readings. I have chosen a number of readings. Please let me know if you have any particular interests so I can try to add an appropriate reading.




next up previous
Next: Three Cautionary Tales

Michael A. Trick
Wed Sep 11 11:04:20 EDT 1996