1. Graphs: MSTs and Shortest Paths David Kauchak cs161 Summer 2009

2. Administrative Grading final grade extra credit TA issues/concerns HW6 shorter will be due Wed. 8/12 before class Errors in class slides and notes Anonymized scores posted

3. Shortest paths What is the shortest path from a to d? A B CE D

4. Shortest paths BFS A B CE D

5. Shortest paths What is the shortest path from a to d? A B CE D 1 1 3 2 2 3 4

6. Shortest paths We can still use BFS A B CE D 1 1 3 2 2 3 4

7. Shortest paths We can still use BFS A B CE D 1 1 3 2 2 3 4 A B CE D

8. Shortest paths We can still use BFS A B CE D

9. Shortest paths What is the problem? A B CE D

10. Shortest paths Running time is dependent on the weights A B C 4 1 2 A B C 200 50 100

11. Shortest paths A B C 200 50 100 A B C

12. Shortest paths A B C

13. A B C

14. A B C Nothing will change as we expand the frontier until we’ve gone out 100 levels

15. Dijkstra’s algorithm

17. prev keeps track of the shortest path

18. Dijkstra’s algorithm

21. Single source shortest paths All of the shortest path algorithms we’ll look at today are call “single source shortest paths” algorithms Why?

22. A B CE D 1 1 3 3 2 1 4

23. A B CE D 1 1 3 3 2 1 4    

24. A B CE D 1 1 3 3 2 1 4    0 Heap A 0 B  C  D  E 

25. A B CE D 1 1 3 3 2 1 4    0 Heap B  C  D  E 

26. A B CE D 1 1 3 3 2 1 4    0 Heap B  C  D  E 

27. A B CE D 1 1 3 3 2 1 4   1 0 Heap C 1 B  D  E 

28. A B CE D 1 1 3 3 2 1 4   1 0 Heap C 1 B  D  E 

29. A B CE D 1 1 3 3 2 1 4 3   1 0 Heap C 1 B 3 D  E 

30. A B CE D 1 1 3 3 2 1 4 3   1 0 Heap C 1 B 3 D  E 

31. 3 A B CE D 1 1 3 2 1 4 3   1 0 Heap B 3 D  E 

32. 3 A B CE D 1 1 3 2 1 4 3   1 0 Heap B 3 D  E 

33. 3 A B CE D 1 1 3 2 1 4 3   1 0 Heap B 3 D  E 

34. 3 A B CE D 1 1 3 2 1 4 2   1 0 Heap B 2 D  E 

35. 3 A B CE D 1 1 3 2 1 4 2   1 0 Heap B 2 D  E 

36. 3 A B CE D 1 1 3 2 1 4 2  5 1 0 Heap B 2 E 5 D 

37. 3 A B CE D 1 1 3 2 1 4 2  5 1 0 Heap B 2 E 5 D  Frontier?

38. 3 A B CE D 1 1 3 2 1 4 2  5 1 0 Heap B 2 E 5 D  All nodes reachable from starting node within a given distance

39. 3 A B CE D 1 1 3 2 1 4 25 3 1 0 Heap E 3 D 5

40. 3 A B CE D 1 1 3 2 1 4 25 3 1 0 Heap D 5

41. 3 A B CE D 1 1 3 2 1 4 25 3 1 0 Heap

42. A B CE D 1 1 1 25 3 1 0 3

43. Is Dijkstra’s algorithm correct? Invariant:

44. Is Dijkstra’s algorithm correct? Invariant: For every vertex removed from the heap, dist[v] is the actual shortest distance from s to v

45. Is Dijkstra’s algorithm correct? Invariant: For every vertex removed from the heap, dist[v] is the actual shortest distance from s to v The only time a vertex gets visited is when the distance from s to that vertex is smaller than the distance to any remaining vertex Therefore, there cannot be any other path that hasn’t been visited already that would result in a shorter path

46. Running time?

47. 1 call to MakeHeap

48. Running time? |V| iterations

49. Running time? |V| calls

50. Running time? O(|E|) calls

51. Running time? Depends on the heap implementation 1 MakeHeap|V| ExtractMin|E| DecreaseKeyTotal ArrayO(|V|)O(|V| 2 ) O(|E|) O(|V| 2 ) Bin heapO(|V|)O(|V| log |V|) O(|E| log |V|) O((|V|+|E|) log |V|) O(|E| log |V|)

52. Running time? Depends on the heap implementation 1 MakeHeap|V| ExtractMin|E| DecreaseKeyTotal ArrayO(|V|)O(|V| 2 ) O(|E|) O(|V| 2 ) Bin heapO(|V|)O(|V| log |V|) O(|E| log |V|) O((|V|+|E|) log |V|) O(|E| log |V|) Is this an improvement?If |E| < |V| 2 / log |V|

53. Running time? Depends on the heap implementation 1 MakeHeap|V| ExtractMin|E| DecreaseKeyTotal ArrayO(|V|)O(|V| 2 ) O(|E|) O(|V| 2 ) Bin heapO(|V|)O(|V| log |V|) O(|E| log |V|) O((|V|+|E|) log |V|) Fib heapO(|V|)O(|V| log |V|) O(|E|) O(|V| log |V| + |E|) O(|E| log |V|)

54. What about Dijkstra’s on…? A B C E D 1 1 -10 5 10

55. What about Dijkstra’s on…? A B C E D 1 1 -10 5 10

56. What about Dijkstra’s on…? A B C E D 1 1 5 10 Dijkstra’s algorithm only works for positive edge weights

57. Bounding the distance Another invariant: For each vertex v, dist[v] is an upper bound on the actual shortest distance start of at  only update the value if we find a shorter distance An update procedure

58. Can we ever go wrong applying this update rule? We can apply this rule as many times as we want and will never underestimate dist[v] When will dist[v] be right? If u is along the shortest path to v and dist[u] is correct

59. dist[v] will be right if u is along the shortest path to v and dist[u] is correct Consider the shortest path from s to v sp1p1 vp2p2 p3p3 pkpk

60. dist[v] will be right if u is along the shortest path to v and dist[u] is correct What happens if we update all of the vertices with the above update? sp1p1 vp2p2 p3p3 pkpk

61. dist[v] will be right if u is along the shortest path to v and dist[u] is correct What happens if we update all of the vertices with the above update? sp1p1 vp2p2 p3p3 pkpk correct

62. dist[v] will be right if u is along the shortest path to v and dist[u] is correct What happens if we update all of the vertices with the above update? sp1p1 vp2p2 p3p3 pkpk correct

63. dist[v] will be right if u is along the shortest path to v and dist[u] is correct Does the order that we update the vertices matter? sp1p1 vp2p2 p3p3 pkpk correct

64. dist[v] will be right if u is along the shortest path to v and dist[u] is correct How many times do we have to do this for vertex p i to have the correct shortest path from s? i times sp1p1 vp2p2 p3p3 pkpk

65. dist[v] will be right if u is along the shortest path to v and dist[u] is correct How many times do we have to do this for vertex p i to have the correct shortest path from s? i times sp1p1 vp2p2 p3p3 pkpk correct

66. dist[v] will be right if u is along the shortest path to v and dist[u] is correct How many times do we have to do this for vertex p i to have the correct shortest path from s? i times sp1p1 vp2p2 p3p3 pkpk correct

67. dist[v] will be right if u is along the shortest path to v and dist[u] is correct How many times do we have to do this for vertex p i to have the correct shortest path from s? i times sp1p1 vp2p2 p3p3 pkpk correct

68. dist[v] will be right if u is along the shortest path to v and dist[u] is correct How many times do we have to do this for vertex p i to have the correct shortest path from s? i times sp1p1 vp2p2 p3p3 pkpk correct …

69. dist[v] will be right if u is along the shortest path to v and dist[u] is correct What is the longest (vetex-wise) the path from s to any node v can be? |V| - 1 edges/vertices sp1p1 vp2p2 p3p3 pkpk correct …

70. Bellman-Ford algorithm

71. Initialize all the distances

72. Bellman-Ford algorithm iterate over all edges/vertices and apply update rule

73. Bellman-Ford algorithm

74. check for negative cycles

75. Negative cycles A B C E D 1 1 -10 5 10 3 What is the shortest path from a to e?

76. Bellman-Ford algorithm

77. G S F E A D B C 10 8 1 3 1 1 2 -2 -4 How many edges is the shortest path from s to: A:

78. Bellman-Ford algorithm G S F E A D B C 10 8 1 3 1 1 2 -2 -4 How many edges is the shortest path from s to: A:3

79. Bellman-Ford algorithm G S F E A D B C 10 8 1 3 1 1 2 -2 -4 How many edges is the shortest path from s to: A:3 B:

80. Bellman-Ford algorithm G S F E A D B C 10 8 1 3 1 1 2 -2 -4 How many edges is the shortest path from s to: A:3 B:5

81. Bellman-Ford algorithm G S F E A D B C 10 8 1 3 1 1 2 -2 -4 How many edges is the shortest path from s to: A:3 B:5 D:

82. Bellman-Ford algorithm G S F E A D B C 10 8 1 3 1 1 2 -2 -4 How many edges is the shortest path from s to: A:3 B:5 D:7

83. Bellman-Ford algorithm G S F E A D B C 10 8 1 3 1 1 2 -2 -4 0        Iteration: 0

84. Bellman-Ford algorithm G S F E A D B C 10 8 1 3 1 1 2 -2 -4 010      8 Iteration: 1

85. Bellman-Ford algorithm G S F E A D B C 10 8 1 3 1 1 2 -2 -4 010    12 9 8 Iteration: 2

86. Bellman-Ford algorithm G S F E A D B C 10 8 1 3 1 1 2 -2 -4 05 10   8 9 8 Iteration: 3 A has the correct distance and path

87. Bellman-Ford algorithm G S F E A D B C 10 8 1 3 1 1 2 -2 -4 05 6 11  7 9 8 Iteration: 4

88. Bellman-Ford algorithm G S F E A D B C 10 8 1 3 1 1 2 -2 -4 05 5 7 14 7 9 8 Iteration: 5 B has the correct distance and path

89. Bellman-Ford algorithm G S F E A D B C 10 8 1 3 1 1 2 -2 -4 05 5 6 10 7 9 8 Iteration: 6

90. Bellman-Ford algorithm G S F E A D B C 10 8 1 3 1 1 2 -2 -4 05 5 6 9 7 9 8 Iteration: 7 D (and all other nodes) have the correct distance and path

91. Correctness of Bellman-Ford Loop invariant:

92. Correctness of Bellman-Ford Loop invariant: After iteration i, all vertices with shortest paths from s of length i edges or less have correct distances

93. Runtime of Bellman-Ford O(|V| |E|)

94. Runtime of Bellman-Ford Can you modify the algorithm to run faster (in some circumstances)?

95. All pairs shortest paths Simple approach Call Bellman-Ford |V| times O(|V| 2 |E|) Floyd-Warshall – Θ(|V| 3 ) Johnson’s algorithm – O(|V| 2 log |V| + |V| |E|)

96. Minimum spanning trees What is the lowest weight set of edges that connects all vertices of an undirected graph with positive weights Input: An undirected, positive weight graph, G=(V,E) Output: A tree T=(V,E’) where E’  E that minimizes

97. MST example A BD C 4 1 2 3 4 F E 5 4 6 4 A BD C 4 1 2 F E 5 4

98. MSTs Can an MST have a cycle? A BD C 4 1 2 F E 5 4 4

99. MSTs Can an MST have a cycle? A BD C 4 1 2 F E 5 4

100. Applications? Connectivity Networks (e.g. communications) Circuit desing/wiring hub/spoke models (e.g. flights, transportation) Traveling salesman problem?

101. Cuts A cut is a partitioning of the vertices into two sets S and V-S An edges “crosses” the cut if it connects a vertex u  V and v  V-S A BD C 4 1 2 3 4 F E 5 4 6 4

102. Minimum cut property Given a partion S, let edge e be the minimum cost edge that crosses the partition. Every minimum spanning tree contains edge e. SV-S e’ e Consider an MST with edge e’ that is not the minimum edge

103. Minimum cut property Given a partion S, let edge e be the minimum cost edge that crosses the partition. Every minimum spanning tree contains edge e. SV-S e’ e Using e instead of e’, still connects the graph, but produces a tree with smaller weights

104. Algorithm ideas? Given a partion S, let edge e be the minimum cost edge that crosses the partition. Every minimum spanning tree contains edge e.

105. Kruskal’s algorithm A BD C 4 1 2 3 4 F E 5 4 6 4 G MST A BD C F E Add smallest edge that connects two sets not already connected

106. A BD C 4 1 2 3 4 F E 5 4 6 4 G MST A BD C 1 F E Add smallest edge that connects two sets not already connected Kruskal’s algorithm

107. A BD C 4 1 2 3 4 F E 5 4 6 4 G MST A BD C 1 2 F E Add smallest edge that connects two sets not already connected Kruskal’s algorithm

108. A BD C 4 1 2 3 4 F E 5 4 6 4 G MST A BD C 4 1 2 F E Kruskal’s algorithm Add smallest edge that connects two sets not already connected

109. A BD C 4 1 2 3 4 F E 5 4 6 4 G MST A BD C 4 1 2 F E 4 Kruskal’s algorithm Add smallest edge that connects two sets not already connected

110. A BD C 4 1 2 3 4 F E 5 4 6 4 G MST A BD C 4 1 2 F E 5 4 Kruskal’s algorithm Add smallest edge that connects two sets not already connected

111. Correctness of Kruskal’s Never adds an edge that connects already connected vertices Always adds lowest cost edge to connect two sets. By min cut property, that edge must be part of the MST

112. Running time of Kruskal’s |V| calls to MakeSet

113. Running time of Kruskal’s O(|E| log |E|)

114. Running time of Kruskal’s 2 |E| calls to FindSet

115. Running time of Kruskal’s |V| calls to Union

116. Running time of Kruskal’s Disjoint set data structure O(|E| log |E|) + MakeSetFindSet |E| calls Union |V| calls Total Linked lists |V| O(|V| |E|) |V|O(|V||E| + |E| log |E|) O(|V| |E|) Linked lists + heuristics |V| O(|E| log |V|) |V|O(|E| log |V|+ |E| log |E|) O(|E| log |E| )

117. Prim’s algorithm

120. Start at some root node and build out the MST by adding the lowest weighted edge at the frontier

121. Prim’s A BD C 4 1 2 3 4 F E 5 4 6 4 MST A BD C F E

122. Prim’s A BD C 4 1 2 3 4 F E 5 4 6 4 MST A BD C F E    0

123. Prim’s A BD C 4 1 2 3 4 F E 5 4 6 4 MST A BD C F E  4 5  60

124. Prim’s A BD C 4 1 2 3 4 F E 5 4 6 4 MST A BD C F E  4 5  60

125. Prim’s A BD C 4 1 2 3 4 F E 5 4 6 4 MST A BD C F E 14 5 420

126. Prim’s A BD C 4 1 2 3 4 F E 5 4 6 4 MST A BD C F E 14 5 420

127. Prim’s A BD C 4 1 2 3 4 F E 5 4 6 4 MST A BD C F E 14 5 420

128. Prim’s A BD C 4 1 2 3 4 F E 5 4 6 4 MST A BD C F E 14 5 420

129. Prim’s A BD C 4 1 2 3 4 F E 5 4 6 4 MST A BD C F E 14 5 420

130. Prim’s A BD C 4 1 2 3 4 F E 5 4 6 4 MST A BD C F E 14 5 420

131. Prim’s A BD C 4 1 2 3 4 F E 5 4 6 4 MST A BD C F E 14 5 420

132. Prim’s A BD C 4 1 2 3 4 F E 5 4 6 4 MST A BD C F E 14 5 420

133. Correctness of Prim’s? Can we use the min-cut property? Given a partion S, let edge e be the minimum cost edge that crosses the partition. Every minimum spanning tree contains edge e. Let S be the set of vertices visited so far The only time we add a new edge is if it’s the lowest weight edge from S to V-S

134. Running time of Prim’s Θ(|V|)

135. Running time of Prim’s Θ(|V|)

136. Running time of Prim’s |V| calls to Extract-Min

137. Running time of Prim’s |E| calls to Decrease-Key

138. Running time of Prim’s Same as Dijksta’s algorithm 1 MakeHeap|V| ExtractMin|E| DecreaseKeyTotal ArrayO(|V|)O(|V| 2 ) O(|E|) O(|V| 2 ) Bin heapO(|V|)O(|V| log |V|) O(|E| log |V|) O((|V|+|E|) log |V|) Fib heapO(|V|)O(|V| log |V|) O(|E|) O(|V| log |V| + |E|) O(|E| log |V|) Kruskal’s: O(|E| log |E| )