Archives
Saturday, April 28. The P4 median was about 17.
Thursday, April 26. We will give a practice final Monday, May 6. (Time and place TBA). We are willing to grade and "count" scores for students who were below the C cutoff for Prelim 1 (= 50) or below the C cutoff for Prelim 2 (= 40). "Count" means that we'll bring those scores up to the cutoff value if the practice final scores warrant.
Thursday, April 26. The grading criteria for Prelim 2, Question 4 have now been changed; we have decided to award SOME credit as appropriate for the following types of responses on this question: 1. On (b), you identified a case when the floating point spacing upon termination is such that you don't end up with exactly the midpoint of the interval. 2. Finding a flaw in (c) that was a flaw outlined in (a) or (b) IF you received a 0 on (a) or (b) respectively. 3. Permuted responses: for example, turning around the answers to (a) and (b). If you gave any of the above responses for this question, you will be given appropriate credit IF you submit a regrade request by 4 PM on MONDAY, APRIL 29. Please do NOT submit regrade requests if you gave other types of responses on this question; they will not be entertained!
Wednesday, April 17. P5 is ready. So is Prelim 2 Solution Guide (PS/PDF). The make-up will be ready to return tomorrow..
Wednesday, April 17. Steve will have exam/P4 pick-up in Upson 4161 from 3-4. More possibilities tomorrow.
Wednesday, April 10. P4 solutions are available. Partial credit guide later.
Wednesday, April 10. Prelim 2 review sheet solutions are available. (PDF/PS).
Monday, April 8. Once your matrix F is set up, the following can be used to set up the normal equations C*alpha = d where C = F'*F and d = F'*y:
Fs = sparse(F);
C = full(Fs'*Fs);
d = full(Fs'*y);
Monday, April 8. Review problems available. (PDF or PS). See you Wednesday nite.
Wednesday April 3. Exam next week. Check details and what to do if you have a conflict. Review problems posted soon.
Wednesday, April 3. For p4, the UTRiSolBand script had a typo which is now corrected. FYI:
b(r:j-1) = b(r:j-1) - x(j)*U(r:j-1,j);
not
b(j-p:j-1) = b(j-p:j-1) - x(j)*U(j-p:j-1,j);
Friday, March 29. Questions about course grading have come up now that the first prelim is over and graded. If you feel that the exam did not reflect what you know it is possible to "recover." When the course is graded a single number is produced according to how things are weighted. But if your prelim grades and final exam grade look like (X,Y,Z) with X<<Y and X<<Z and the computing assignments reflect a strong effort then I will downplay the contribution of Prelim 1 to your final grade. No formulas here, just a statement that grading is not mechanical and that you are given the benefit of the doubt if a case can be made as in my XYZ example. With respect to the computing assignments, be aware that the scores for the assignments are much higher and more uniform than for the exams and that they more or less amount to adding a constant to everybody's exam summation.
Friday, March 29. Make-up prelim folks: a solution guide is available in 4130. I will bring it to class on Monday.
Thursday, March 28 The P4 scripts and function specifications are available. Be sure to follow the specifications for BEval and LSSpline that are on this website. They differ slightly than what is given in the handout.
Monday, March 25 Prelim1 solution guide available. (PS or PDF). The partial credit scheme is NOW included in these documents. Feel free to submit regrades within the next week. Note that we look over your whole exam even if you just want us to look at a single problem. Scores can go up or down. So carefully compare your solutions with our solutions. Exams returned via TAs during their office hours. Note that tomorrow Mike and Suzanne are adjusting their hours so as to be available from 10:30-3 in Rhodes 657. Grading guide for make-up will be brought to class on Wednesday.
Monday, March 11. The make-up prelim is Wednesday, March 13, 5-6:30 pm, Upson 205.
Monday, March 11. Review sheet solutions and P3 solutions are available.
Tuesday, March 5. Next week is Prelim 1 week. Be aware of where to go and when.
Tuesday March 5. Prelim 1 review problems are available. (PS or PDF). Hardcopies will be brought to lecture.
Monday, March 4. QUAD does composite Simpson rule with the subintervals determined adaptively. In P3A, you are to estimate the sector areas using (simple) Simpson's rule. Thus, QUAD is not suitable for computing the sector areas.
Monday, March 4. QUAD is recursive and Matlab places a limit on the maximum depth of the recursion. For the tolerances that you are to try, this limit should not be reached. (At least it wasn't on my system.) However, if you are having a problem with this in P3B, then try QUAD8. This is based on a higher order Newton Cotes rule and (hopefully) will not require so many recursive calls. The calling sequence for QUAD8 is the same as for QUAD.
Monday, March 4. In the second bullet, page 2, n = nSect and nSect is defined in the first bullet.
Friday March 1. Prelim coming up on Tuesday, March 12. Here is the Prelim 1 solution guide from the Spring 2000 version of the course. More review problems next week.
Friday, March 1. At the end of the write-up of Part A, change "12" to "13". The script P3A will draw 13 sectors if your functions are correct so this will harmonize with P3a and also the figure that is in the handout. Also, you can vectorize your Simpson estimation of the sector-areas. Just because we use "for loop language" in describing this part of the algorithm doesn't mean that you should use for-loops in your implementation.
Friday, March 1. Students are asking about how the scores are running on P1 and P2. The median for those two assignments were 15 and 17 respectively (out of 20).
Wednesday, February 27. The syllabus and section assignments have been permuted a bit.
Monday, February 25. P2 solution is available. P3 handout and scripts are available. P2 and P1 will be handed out in lecture this wednesday. There is an exam in 2 weeks.
Tuesday, February 19. Change in Section problems for this week: P4.2.2, P4.3.2, and P4.3.6.
Thursday, February 14. Lecture examples from this week and last. And the solutions to P1.
Monday, Feb 11. Clarifying points for P2. (1) That is exp(-t/10) = e^{-t/10}. (2) All mentions of spline in this assignment are the "not-a-knot" spline that the Matlab spline function returns, i.e., S = spline(x,y) with length(x) = length(y). (3) Your function SunSpotSpline does not have to check that Y1<=Y2 and nSmooth>=0. (4) In Part C the SunSpotSpline will have local maximums and local minimums. It is these that you must compute. We are NOT interested in the global maximum (or minimum) in the interval [1900,2000].
Thursday, Feb 7. Assignment P2 is ready. We'll bring hardcopies to the sections and the lecture next week. It is about estimating the time in between sunspot cycle peaks.
Wednesday, Jan 30. I will post most Matlab scripts/functions that I use in lecture. Take a look at what came up this week.
Wednesday, Jan 30. The posted office hours for TAs and CVL are walk-in hours. No need for appointment. No need to sign up in Upson 303.
Thursday, Jan 24. In the P1 handout on page 3, MakeEllipse should return the empty matrix if A' <= 0, C'<=0 or Ftilde<=0. (See the line after the specification of this function.)
Thursday, Jan 24. A newsgroup for the course has been established. To subscribe to the newsgroup, select the Newsgroups option from the Communicator menu within Netscape, then double click on newsstand.cit.cornell.edu and subscribe to cornell.class.cs322. Use theGet Msg and New Msg/Reply icons to check for and post messages,respectively. You will be prompted for your netid and password when posting a message. Note, we are not sure if the TAs have enough time to monitor the newsgroup. This will become clear in a week or two and we will let you know.
Thursday, Jan 24. The P1 Due date is Wednesday, February 6. The handout said Wednesday February 9.
Wednesday, Jan 23. The P1 handout has typos in equation (2). There is a missing x and a missing y inside the big parentheses and it should be C' not just C.