2. A talk with 3 titles By Patrick Prosser

3. Research … how not to do it LDS revisited (aka Chinese whispers) Yet Another Flawed Talk by Patrick Prosser

5. Send reinforcements. We’re going to advance.

7. Send three and fourpence. We’re going to a dance!

8. A refresher Chronological Backtracking (BT) what’s that then? when/why do we need it? Quick Intro Limited Discrepancy Search (lds) what’s that then Then the story … how not to do it

9. An example problem (to show chronological backtracking (BT)) Colour each of the 5 nodes, such that if they are adjacent, they take different colours 12 3 4 5

10. A Tree Trace of BT (assume domain ordered {R,B,G}) 12 3 4 5 v1 v2 v3 v4 v5

11. A Tree Trace of BT (assume domain ordered {R,B,G}) 1 2 3 4 5 v1 v2 v3 v4 v5

12. A Tree Trace of BT (assume domain ordered {R,B,G}) 12 3 4 5 v1 v2 v3 v4 v5

13. A Tree Trace of BT (assume domain ordered {R,B,G}) 12 3 4 5 v1 v2 v3 v4 v5

14. A Tree Trace of BT (assume domain ordered {R,B,G}) v1 v2 v3 v4 v5 12 3 4 5

15. A Tree Trace of BT (assume domain ordered {R,B,G}) v1 v2 v3 v4 v5 12 3 4 5

16. A Tree Trace of BT (assume domain ordered {R,B,G}) v1 v2 v3 v4 v5 12 3 4 5

17. A Tree Trace of BT (assume domain ordered {R,B,G}) v1 v2 v3 v4 v5 12 3 4 5

18. A Tree Trace of BT (assume domain ordered {R,B,G}) v1 v2 v3 v4 v5 12 3 4 5

19. A Tree Trace of BT (assume domain ordered {R,B,G}) v1 v2 v3 v4 v5 12 3 4 5

20. A Tree Trace of BT (assume domain ordered {R,B,G}) v1 v2 v3 v4 v5 12 3 4 5

21. A Tree Trace of BT (assume domain ordered {R,B,G}) v1 v2 v3 v4 v5 12 3 4 5

22. A Tree Trace of BT (assume domain ordered {R,B,G}) v1 v2 v3 v4 v5 12 3 4 5

23. Could do better Improvements: when colouring a vertex with colour X remove X from the palette of adjacent vertices when selecting a vertex to colour choose the vertex with the smallest palette tie break on adjacency with uncoloured vertices An inferencing step A heuristic (Brelaz) Conjecture: our heuristic is more reliable as we get deeper in search

24. What’s a heuristic?

25. Limited Discrepancy Search (LDS)

27. Motivation for lds

28. Motivation for LDS

29. LDS show the search process assume binary branching assume we have 4 variables only assume variables have 2 values each Limited Discrepancy Search

30. Limited Discrepancy Search (LDS)Ginsberg & Harvey Take no discrepancies (go with the heuristic, go left!)

31. Limited Discrepancy Search (LDS)Ginsberg & Harvey Take no discrepancies

32. Limited Discrepancy Search (LDS)Ginsberg & Harvey Take no discrepancies

33. Limited Discrepancy Search (LDS)Ginsberg & Harvey Take no discrepancies

34. Limited Discrepancy Search (LDS)Ginsberg & Harvey Take 1 discrepancy

35. Limited Discrepancy Search (LDS)Ginsberg & Harvey Take 1 discrepancy

36. Limited Discrepancy Search (LDS)Ginsberg & Harvey Take 1 discrepancy

37. Limited Discrepancy Search (LDS)Ginsberg & Harvey Take 1 discrepancy

38. Limited Discrepancy Search (LDS)Ginsberg & Harvey Take 1 discrepancy

39. Limited Discrepancy Search (LDS)Ginsberg & Harvey Take 1 discrepancy

40. Limited Discrepancy Search (LDS)Ginsberg & Harvey Take 1 discrepancy

41. Limited Discrepancy Search (LDS)Ginsberg & Harvey Take 1 discrepancy

42. Limited Discrepancy Search (LDS)Ginsberg & Harvey Take 1 discrepancy

43. Limited Discrepancy Search (LDS)Ginsberg & Harvey Take 1 discrepancy

44. Limited Discrepancy Search (LDS)Ginsberg & Harvey Take 1 discrepancy

45. Limited Discrepancy Search (LDS)Ginsberg & Harvey Take 1 discrepancy

46. Now take 2 discrepancies

47. Limited Discrepancy Search (LDS)Ginsberg & Harvey Take 2 discrepancies

48. Limited Discrepancy Search (LDS)Ginsberg & Harvey Take 2 discrepancies

49. Limited Discrepancy Search (LDS)Ginsberg & Harvey Take 2 discrepancies

50. First proposal For discrepancies 0 to n

51. First proposal For discrepancies 0 to n k is remaining discrepancies

52. First proposal For discrepancies 0 to n k is remaining discrepancies Go with heuristic

53. First proposal For discrepancies 0 to n k is remaining discrepancies Go with heuristic Go against then go with

54. The lds search process: how it goes (a cartoon)

55. The lds search process: how it goes NOTE: lds revisits search states with k discrepancies When searching with > k discrepancies

56. My pseudo code

57. lds revisits nodes: Korf’s improvement (AAAI 96)

58. Korf’s improvement

59. Korf’s 1 st mistake! Woops! Do you see it? He’s taking his discrepancies late/deep!

60. Wrong way round. Is that important? Harvey & Ginsberg Korf

61. Korf’s 1 st mistake! Wrong way round Richard

62. Richard, was that a bug?

63. Yes, but so?

64. Korf’s 2 nd bug

65. Woops! Richard, you know there is another bug?

66. My pseudo code

67. Has anyone noticed Korf’s bug? Have people been using Korf’s LDS? Have people been using Harvey & Ginsberg’s LDS? Has anyone remembered the motivation for LDS?

71. Chris, late or early?

72. Wafa, late or early?

73. Wafa’s response

74. My pseudo code I think this has not been reported

75. Does it make a difference if we take discrepancies late or early? An empirical study Tests Harvey & Ginsberg’s motivation for LDS

76. Car Sequencing Problem Assessed exercise 2

88. My empirical study on car sequencing problems Using various search algorithms, heuristics. Question: does the order (late/early) that we take discrepancies in lds matter? is the order sensitive to the heuristics used?

89. Performing the experiments (what’s involved) code up lds in JChoco for non-binary domains paramaterised late/early discrepancies using Korf’s improvement code up model of car sequencing problem using Pascal Van Hentenryk’s model code up my BT (as a gold standard) code up a certificate checker is a solution a solution? code up 4 different heuristics 2 published heuristics for car sequencing the 2 anti-heuristics Perform experiments on benchmark problems limits on CPU time (minutes sometimes hours per instance) test that all solutions are solutions (paranoia?) problems typically have 200 cars (non-trivial) NOTE TO SELF also did Golomb rulers started on HC did this to show results were general and not car seqn specific

93. and now the results …

95. Well, did you see a pattern? If there is no pattern what does this say about H&G’s hypothesis? And, if no pattern, why is lds any good?

96. See anything?

97. Got my act together for ECAI08 reject

98. How about another problem domain? Hamiltonian Circuit ecai08 rejects

100. What’s involved? ecai08 rejects

103. HCP: What’s involved? constraint model, single successor subtour elimination constraint heuristic constraint model maintains information resultant model is then Algorithm 595 (a surprise) locate benchmark problems perform experiments late v early discrepancies heuristics static v dynamic ecai08 rejects

105. Still to do lds - extremely early and extremely late jobshop scheduling using Cheng & Smith’s heuristic repeat H & G’s experiments going late/early number partitioning with CKK repeat Korf’s experiments going late/early fair bit of implementation and analysis 2 months work at least should get 2 more tables might get new ****09 rejection 

106. What lessons can we learn? we can just follow on without question (read the most recent paper) forget the basic/initial hypothesis forget to really look at our results take too much for granted it is not uninteresting to repeat someone elses’s experiments. do not be frightened or disinterested in –ve results or different results (above) don’t do just one set of experiments when you can do many (different domains) published papers may have multiple errors (beware) be paranoid (Am I right? Is he right? How do I know I’m right?) know that we are human

107. Latest reject Available from the kiosk ecai08 rejects

108. Start without me Questions?