APPM 6640, Spring 2009, Coursework

COURSEWORK

for APPM 6640

HOMEWORK

Due one week after relevant chapter is covered. Late homework will not be accepted!
Some solution sets include more problems than were assigned in case you're interested.

Chapter 1:   1, 3,4   Due January 21st!
      HW1 Solutions
Chapter 2:   1-3, 8, 10, 13, 16   Due February 6th!
      HW2 Solutions
Chapter 3:   2-6   1-D only!   Due February 18th!
      HW3 Solutions
Chapter 4:   6, 7, 11   Due March 6th!
      HW4 Solutions
Chapter 5:   2, 13   Due March 13th!
      HW5 Solutions
Chapter 6:   1, 2, 6
      HW6 Solutions
         As listed in the book's errata,
         there should be a minus sign
         in front of u" in 6 (i) & (ii).
Chapter 7:   1, 2, 10, 12, 15, 21
      HW7 Solutions

We probably won't assign homework
beyond chapter 7, but just in case:

Chapter 8:   2, 6, 8
Chapter 9:   1, 3
Chapter 10:   1, 5, 8, 9

Please submit the homework through Dropbox . Instructions for doing this will be given during the first computing lab. It would help us greatly if you could make an account before the first computing lab.

PROJECT

The team project involves individual groups working together to study some aspect of multilevel methodology. The choice of which aspect to pursue, how to pursue it, and how to present the results is up to the team. The major role of the instructors is as facilitators. They can be used as resources for information, ideas, guidance, and general assistance, but the more initiative that the team exhibits, the better.

* Group into teams of possible common interest.
 We will discuss in class possible areas of interest, then break into a few teams.
* Meet in lab (typically every other session).
 Approximately every other lab period will be at least partly devoted to team meetings.
* Identify an application area of interest.
 The first task is to identify a general topic. Discuss your interests and be sure that everyone is comfortable with what you choose. Don't obsess too much how applicable the topic is. You can start researching the area and adjust as you go. No need to fixate on your first choice if something better comes along.
* Identify a problem in that area & learn about it.
 Start reading journal articles and books and such. This campus has a lot of material and personnel resources. Use them. Find out more about the topic and what its burning problems are. Choose one that hopefully gives you a chance to begin to tackle. Think in terms of being able to identify a very very simple version of that problem first.
* Learn current methods & their limitations.
 How is that field approaching the problem now? What bothers the researchers. What would they really want to do if they had better methods? Can you see what's holding the field back on this problem?
* Get experience with these methods on typical cases.
 Try their methods out on a problem--however simplified you need to make it to proceed. Play around with it to see how it performs over a range of parameters, input, and demands.
* Try cases where these methods (begin to) fail.
 Do you see the limitations of these methods? Can you up your demands (larger resolution, better accuracy, more features) to begin to stall or blow up the computations?
* Brainstorm a better (multilevel?) method.
 This is the crux: think about what you're learning in this class and whether it gives you any insight as to how to improve on the current methods for your problem. If thinking fails, just take a stab at an approach. It's better to have an idea based on intuition or analysis that suggests it will work, but lack of insight shouldn't prevent you from trying something.
* Implement & test your idea.
 Give your idea a shot. Implement and test it on your problem, starting with the simplest possible version first and introducing more difficulty as you proceed. Get as far as you can in the time allowed in this course.
* Inform the rest of the class along the way.
 From time to time, one of your team members should take a few moments to clue the rest of us in on your progress. Just a short summary of where you are and where you're heading should suffice.

What happens if you don't get very far with your research? What happens if you have no idea for a better method? What if your idea goes down in flames? Well, unfortunately, there's no way to predict how your project will go. The old saying is: "If we knew what we're doing, we wouldn't call it research." So, if you "fail", then welcome to the club. That's just part of the process. So, the answer is, don't worry. Just do your best. That's all we can ask for. There's much to be learned in an "unsuccessful" project too.

Project Report Suggestions

Your team should document your experience by writing a short report, much like a journal paper. It should have a title, alphabetically ordered list of authors (your team members), abstract, introduction, and several sections that typically include introduction, problem statement, methodology, numerical results, and conclusions. Along the way as your team conducts research, pay attention to how journal papers are written. Then, starting possibly as early as mid semester, write a skeleton report and begin to fill it in as your research develops. Remember, graphs are most illuminating for the reader, so please include them in addition to or possibly even in place of big tables.

Project Presentation Suggestions

Each of your team members should participate in presenting your results to the class--at the end of the semester. When you make that presentation, remember that you are speaking to those who are unfamiliar with your work. Be simple, clear, and precise. Let us know what you learned, give us an indication of how much effort you put into it, and suggest what you would've done if you'd had months more to continue on the project. What fundamental principles would you like to impart to us that came to light? Your observations and opinions are what counts most in your project. Put yourself in the presentation. Please give us a glimpse as to who did what. We should end up with an idea of how each of you contributed individually to the effort.

Using a beamer and a PowerPoint or PDF or whatever presentation might be a good idea. It will save you time and taxation of your memory. Put just a few lines on each slide, with just a few phrases. Do NOT show us lots of numbers, long sentences, or many equations. Keep it clear and simple, with just one or two main points per slide.

Practice, practice, and then practice. Time your talk, but expect it to go slower than in practice. Out of curiosity, we will hopefully ask lots of questions.

Most of all, relax and have fun with this.