The Ph.D. Qualifying Exam in ECE at Carnegie Mellon

(Revision 1.01)

Last updated 16 October 1997

Disclaimer: The contents of this page are merely the opinions of the author. They are not necessarily the opinions or policies of the ECE department. While generalizations about the exam can be made, each exam is unique, and the generalizations may not be true for a particular exam. While I think the advice offered will help you prepare for the exam, there can be no guarantee of success.

The Ph.D. qualifying exam consists of the student giving a 20 minute presentation to a qualifying committee of three professors, followed by the committee grilling the student for up to three hours. The committee is made up of one professor from the area of the student's presentation, one in an area closely related to the student's presentation, and one from an area unrelated to the presentation.

For example, if a student were talking about computer architecture, the first committee member would work in computer architecture, the second might be in CAD or signal processing, and the third in electromagnetic fields.

The committee is arranged in this fashion to test both the breadth and depth of the student's knowledge.

Dan's Qualisms

When I was preparing for the qual, my advisor, Dan Siewiorek, gave me a list of "qualisms", little nuggets of information on what can happen during the qual. Here are some of them with commentary from me on their meaning.

• Beware of the garden path--someone can feed you little bits that individually seem logical, but lead you to an indefensible position.
  My commentary: Imagine that you are walking in a garden near a swamp with a guide. You do not want to encounter any alligators. The guide points to the first step in the garden, showing you that there are no alligators on it. So you step into the garden and admire the beautiful swamp flowers. When the guide urges you to move on, you look at the second step--no alligators. So you go to the second step and admire the cypress stumps. At the third step there are puddles of water, and the trees have changed, but you see no alligators. The guide tells you about how the flora and fauna changes rapidly near the swamp as the height of the ground changes from a few feet above sea level to near sea level. You find this very interesting, and while you note that there is more water at each step, you do not see any alligators. A few steps later, the guide asks you to pull a floating vine from the water so that you may examine it more closely. You grab the vine and suddenly realize that it is no vine--you have an alligator by the tail!
  
• Be careful of saying, "I know my answer because I did the work." The committee may take your problem formulation and change the assumptions.
  My commentary: The committee will try to see if you can take what you've learned on one problem and apply it to another problem. If the assumptions of your problem change, the results may change as well. Don't automatically stick to your answers to the original problem. See if the change in assumptions will effect the results.
  
• Articulate. The committee wants to know what you're thinking.
• Explore the boundaries of the problem by using degenerate cases.
• The committee is sitting back doing pattern matching for classical EE.
• Two things you have to do: Talk and write. Each buys you time to think.
• Summarize the problem to make sure you understand the question correctly.
  My commentary: If the question you answer is not the question that was asked, you run two risks: Looking stupid (for lack of a more tactful term) and opening up a line of questioning that the committee did not previously have in mind.
  
• Write down the assumptions you are making to avoid any confusion.

Dan's fundamentals for computer engineering students

• Data and Control path
• ISA (know at least one)
• Cache, TLB orgnization
• Bus protocols
• Network protocols
• Latency vs throughput (latency vs bandwidth)
  -Being able to explain all the pieces on a diagram of a CPU with its memory hierarchy
  -Explain why particular features impact latency
• Know both software and hardware layers
  -For instance: Type command "ls" from keyboard on a Unix shell until the the characters and output appear on the screen, what are involved at different layers: device, driver, disk storage, memory, printer buffer, display ...
  -Email: network (protocol stack, header info, error correction, destination, name server ...)

• RC circuits using the differential equations
• Concurrency, i.e. critical sections, race conditions, deadlock, semaphores, lock variables, monitors, mutual exclusion.
  -Be able to explain the concurrency involved in a simple situation, two users printing at the same time, for example.

My advice on studying

You do not want to be nervous during your qual. The best way to not be nervous is to know that you have prepared as well as possible and to be rested.

Form a qual study group about six months before you plan to take the qual. Use the qual group to study engineering fundamentals that you have probably forgotten through disuse: calculus, differential equations, probability and statistics, circuit analysis. The qual group will probably be made up of students from each area of ECE, but all of you will need to know these fundamentals. As the time for your qual grows nearer, have "mock quals" with your qual group: Let several of them act as your committee, give your talk, and answer their questions. Mock quals with students of your committee members is also a good idea.

About two or three months before the qual, decide what areas are related to your talk, pick the best textbooks in those areas, and work out a reasonable schedule to cover the material. Keep track of your progress so that you don't fall behind. One can easily get bogged down when working alone. If you get stuck trying to understand something, go on to another topic and come back to it later.

At about the same time, you'll have to write the abstract which gets submitted to the department. Write the abstract carefully: It determines who gets put on your committee and what questions they'll think of ahead of time.

It will be difficult to schedule a time for your exam. (An ECE folk saying: The hardest part of preparing for your qual is finding a time to have it.) As soon as you find out who your committee is, start trying to schedule the exam. The summer is an especially bad time for scheduling a qual because of travel. In my case, there were less than 10 days over the whole summer when all three members of the committee were in town. During the fall and spring semesters, don't count on having the exam during finals or the week or two before. Professors are very busy then and will usually ask that you not schedule it for that time. Don't forget to reserve a room for the exam.

Leave the last two weeks before the qual open for working problems and practicing your talk. Do not plan to cover new material in this time. Practice working problems at the chalkboard, and practice talking through the problem as if the committee were in the room. If possible, use the room where you'll be taking your exam.

Look up recent papers by your committee members to find out what their current research interests are. Think of questions someone with those research interests might have when seeing your talk.

The night before the exam is no time to be studying. If you haven't learned the material already, you won't learn it in the last 24 hours. Do something to unwind. I went out for beer and nachos. Another person I know went around test-driving sports cars at dealerships. Be adventurous in your unwinding. Then go home and get a good night's sleep.

Other qual links:

• Pascal Meier's qual advice page: Mainly advice on qual slides.

Acknowledgements

Comments? Send me mail:

Back to my home page