Note on the format of CmpE360 homeworks:

There are some guidelines you should follow for CmpE360 homeworks. Make sure that you follow these to receive full points. Also you’ll realize that this way your homeworks won’t take too much time and effort.

• The code is unimportant. The important thing is what you make out of the assignments. So clearly explain your reasoning, and give the results. That’s what counts.
• Do you understand your results? There were so many obviously wrong answers submitted for assignment 1. This is not a C or Matlab course. Don’t make your code user-friendly or anything. Just make sure that you understand what you are doing. For example, casting the error to integer in the middle of Kahan’s summation method means that you have no idea what you are doing.
• Two pages of a clearly written report are much more worth than ten pages of uncommented code, and they are usually sufficient for weekly assignments.
• Never ever submit just your source, possibly with the exe file, without any kind of explanation. You may entirely omit the code (if not stated otherwise), or even just give pseudo code, but you should always give some kind of report.
• Don’t ignore the ‘sub-questions’ in the homeworks. For example, if, as in HW2 Q1, the conditions are asked, answer it. That’s the easiest way to show that you understood the question.
• The matlab code for HW2 doesn’t take more than a half page. If you find the homeworks too hard, you won’t after you learn Matlab. So make sure that you learn it fast. Nevertheless the upcoming assignments will be shorter.
• Read and think about the homeworks as soon as they are assigned. When you think there is something unclear about a question, don’t hesitate to ask the TA about it. You may use the CmpE360 e-mail list for this purpose (make sure that you register to the list). If there is no time left to do that, clearly state your assumption and then answer the question.
• Finally, don’t forget that CmpE360 questions are usually tricky and/or complicated. Think about the question first. Why is it asked? The answer may be more complicated, or much simpler than you think. Also this way you may realize that you have done something wrong, for instance, if it turns out that Kahan’s method makes worse than single precision in HW1. If so, why is there a need for such a method? Think about the question, think about your answer.