ECS 222A - Fall 2007 Algorithm Design and Analysis - Gusfield

  • This index page will just link to the various course handouts that are available on the web, and provide some description of them.

    Distribution List

    1. Videos for lectures on Graduate Level Algorithms An annotated list of video lectures for CS222A, along with supplemental lectures.

    2. Course Syllabus
    3. Introduction to complexity and rules of the game
    4. HW 1, due October 8. Read about homeworks on the syllabus.
    5. Rant on induction - might be helpful if you are unsure about how to do inductive proofs
    6. Notes on using the Master Method to solve divide and conquer recurrence relations
    7. Different notes on solving recurrence relations and the Master method.
    8. The class TA, Till Stegers, will hold office hours Tues. 3-5; Wed. 3-5; and Thurs 10-12, in Room 53 Kemper Hall. These will start tomorrow (Oct. 2). Gusfield's office hours will be Monday 4-6 and Thurs. 5-6, and by appointment. We will announce information about a possible discussion section in the class on Weds, and on this site.
    9. A Running list of topics covered in lecture
    10. We will have a weekly discussion on Weds. 6-7 (starting tonight Oct. 3) in the same room as the class in Bainer hall. It is not certain that these will be able to be televised, but that is a future possibility.
    11. HW 2, due October 17. The page reference for problem 2 should be 329, not 239.
    12. Weds. class and discussion section will be taught by Till Stegers. Gusfield will not have thursday office hours October. 11.
    13. HW 1 Solutions
    14. The midterm will be Friday October 26.
    15. HW 3, due October 26.
    16. The HW 2 posted above now includes a write up of the way we discussed the interleaving problem in the Oct. 17 discussion section, for people who did not attend.
    17. Homework 1 scores: Max possible 40, Max achieved 36, Average 23.83, Median 25, Average in terms of percent of Max possible 59.58%.
      I was very impressed by how much work people put into this assignment, and how sophisticated some of the answers were. I think the TAs also put a huge effort into grading quickly, making many comments that should help you understand any deficiencies in your solutions. Problem 1b was not graded. We may often need to do that; not grade some problems in order to concentrate on grading others in depth.
      If you have questions about the grading on problem 1, see Till Stegers in his office hours. If you have questions about Problems 2 and 3 on HW 1, hold off for now until - more information forthcoming.
    18. I was asked what the course material is, since the exact topics in the syllabus do not exactly match what we have done. The material of the course is what we have covered in class or homework or any indicated reading specified in a homework. The running list of lecture topics posted above lists the topics we have discussed in class. However, it may be that I can craft a fair question in the midterm about something that wasn't explicitly in the class material,if I think it is close enough to class material, so that a student who understands the class material should be able to figure out that question.
    19. Our second TA Juan Lang will have one weekly office hour to discuss homework problems that he graded. On homework 1 those are problems 2 and 3. He will have office hours on Tuesady from 4 to 5 in 53 Kemper.
    20. HW 2 Solutions
    21. Discussion section Wed. Oct. 24 will be held as usual.
    22. Gusfield's office hour Thursday Oct. 25 is cancelled.
    23. Midterm Solutions
    24. Homework 4
    25. Notes on the sports elimination theorem discussed October 29
    26. Solutions to HW 3 The solution to problem 4 has been modified Nov. 1, 2007.
    27. Till's solution to the Node Cover problem from HW 3
    28. Webcast of the first lecture 9-28-07
    29. Webcast of 10-1-07 - trial
    30. Webcast of 10-3-07 - trial
    31. Webcast of 10-5-07 - trial
    32. Webcast of 10-8-07 - trial
    33. Webcast of 10-10-07 - trial
    34. Webcast of 10-12-07 - trial
    35. Webcast of 10-15-07 - trial
    36. Webcast of 10-17-07 - trial
    37. Webcast of 10-19-07 - trial
    38. Webcast of 10-22-07 - trial
    39. Webcast of 10-24-07 - trial
    40. Webcast of 10-29-07 - trial
    41. Webcast of 10-31-07 - trial
    42. Webcast of 11-2-07 - trial
    43. Webcast of 11-5-07
    44. Webcast of 11-7-07
    45. Webcast of 11-9-07
    46. Webcast of 11-14-07
    47. Webcast of 11-16-07
    48. Webcast of 11-19-07
    49. Webcast of 11-21-07
    50. Webcast of 11-26-07 Some part of the beginning of this lecture was lost. The topic is a deterministic algorithm for global minimum cut in an undirected graph. This algorithm is not discussed in the book, but a randomized algorithm for the problem is discussed, so you can read the start of that discussion for the definition of the problem, and for some other comments on the problem that relate to both the deterministic and the randomized algorithms.
    51. Test video posting
    52. Webcast of 11-28-07
    53. Webcast of 11-30-07
    54. Webcast of 12-3-07
    55. Webcast of 12-5-07
    56. Webcast of 12-7-07 Oops, at the end of the lecture (the end of the course) things got a bit rushed and I wrote that the DP for independent set for graph G runs in time that is polynomial in w(T) and n and m, where w(T) is the width of the tree decomposition T, and n and m are the number of nodes and edges in the graph G. What is correct is that the DP runs in time that is exponential in w(T) (in fact it contributes a factor of 4^{w(T)}), and polynomial in n and m. The actual running time is something like O(4^{w(T)}nm). The point is that if w(T) is bounded, independent of n and m, then 4^{w(T)} is just a constant number, so independent set problem can be solved in polynomial time as a function of n and m (again if w(T) is bounded independent of n and m). Similarly, even if w(T) is not bounded, if it is small, then the running time is practical for realistic n and m. The smaller w(T) is, the more tree-like is the graph, and the faster the DP.
    57. Webcast of 12-11-07

    58. webcast of discussion Oct. 10, 2007
    59. webcast of discussion Oct. 17, 2007
    60. webcast of discussion Oct. 24, 2007
    61. webcast of discussion Oct. 31, 2007
    62. webcast of discussion Nov. 7, 2007
    63. webcast of discussion Nov. 14, 2007
    64. There are no webcasts of discussions between Nov. 14 and Dec. 5, although the videos of those are available.
    65. webcast of Discussion Dec. 5, 2007
    66. distribution of hw scores:
      Assignment: HW1, HW2
      Max possible: 40,30
      Max achieved: 36,29
      Average: 24.05, 21.78
      Average in percent: 60.12%, 72.58%
      Median: 25, 22.5
      Median in percent: 62.50%, 75.00%
    67. Comments on HW 4: Yes problem 1 is easy. Problem 2 asks for a solution using one flow. More exactly, a single network flow computation suffices, after determining some legal assignment. Hence if we want to compute v*(i,j) for each empty cell (i,j), it suffices to compute one flow to determine a first legal assignment, followed by one flow per empty cell.
    68. Midterm grade information:
      Max possible 90
      Average 38.83
      Average in percent 43.14%
      Median 40
      Median in percent 44.44%
      Max 76
      Max in percent 84.44%
      Min 7
    69. Till is moving his thursday (Nov. 8) office hour to 1-3pm, this week only.
    70. Homework 5
    71. NOTE on Nov. 8, 2007. No need to do problem 2 on HW 4. The hint I gave you on how to solve problem 2 on HW 4, i.e., to use the solution I gave in problem 4 of HW 3 was likely confusing, because the second half of problem 4 on HW 3 had an error in it. I am now writing up corrections to problem 4 of HW 3 and will post them here later. But given the added difficulties this probably caused, you may skip problem 2 of HW 4 if you have not already done it. I apologize for the confusion.
    72. Corrected solutions to HW 3 problems, and a discussion of Problem 2 on HW 4.
    73. Several people have asked whether the node cover question in problem 2 of HW 5 is restricted to a tree, as is the greedy independent set method. The answer is yes.
    74. Notes on the Steiner String Problem discussed Nov. 14, 16
    75. Homework 6
    76. There will be NO discussion section on Wed. Nov. 21. Happy Thanksgiving.
    77. Till will not have any more Thursday 10-12 office hours. His other office hours will remain the same.
    78. Solutions to HW 4
    79. Notes on the Z-algorithm
    80. Note on the grading of HW 3. Problem 2 was not graded (to save time and because the problem is similar to problem 1), and the second part of Problem 4 was also not graded.
    81. Homework 7 Corrections made in the periodicity problem Dec. 5
    82. A simple min-cut algorithm Presented in class Nov. 25. This may not be accessable from outside of UCD. The citation is J. of the ACM, Vol 44, No. 4, July 1997, p. 585-591
    83. Solutions to HW 5
    84. Nov. 30 and Dec. 3 we discussed a simple algorithm for the Constant-Time LCA problem. I was following unpublished notes titles ``The LCA Problem Revisited" by M. Bender and M. Farach-Colton. I notice that many people on the web have posted those notes, so you can find them through google. Being more mindfull of copyright, I will not post those notes here. The notes seem also to be posted on Benders homepage www.cs.sunysb.edu/~bender/ but I couldn't get those to download properly. There also is a more complete and complex published version of that paper.
    85. The first part of one of the Final Exam questions
    86. The following fact may be very useful for the periodic string problem on HW 7: If a string is periodic with period A and also with period B, then it is also periodic with period GCD(A,B). In particular, A and B are multiples of GCD(A,B). You can use this fact in the HW without needing to prove it. Also, there was a typo in the original posting of this problem, and it has been corrected along with some other changes, and HW 7 has been reposted.
    87. December 6, 2007. Sorry I (Gusfield) will not be in my office today for office hours. I will be back in town tomorrow (friday), so send me mail if you want an appointment. I will hold normal office hours next week on monday, and will be available by appointment.
    88. Solutions to HW 6
    89. Solutions to HW 7 I am off by 1 in several places in the Solution to Problem 2, but the ideas are basically correct. Draw a picture of a periodic prefix to get the idea of the solution, and from there you can figure out exactly what the indices should be.
    90. Graded Homework 5 papers are in a box outside of my office door - 2125 Kemper.
    91. The rules for the final exam are the same as for the midterm - open book and notes, but only the two books associated with the class, and only your own notes and the notes/papers distributed through the class. In fact, some problems may assume that you have the book with you, i.e. refering to definitions you have seen, but not rewriting them. Also, it would help again if you brought a small blue book - much easier to keep the papers together, reducing the probability that yours will be lost.
    92. Graded homeworks 6 and 7 are outside the door at 2125 Kemper. On HW 7, problem 2 and problem 4b were not graded.
    93. Most people missed the point of question 4a on Homework 6, and thought it was a trivial question. Some have come to talk to me about it. I have now written a full explanation of the problem with the trivial solution, and what a correct solution should be.
      Solutions to HW 6 with expanded discussion of problem 4a