1 / 32

DAA U-2(part1)

DAA disjoint sets

Srinu8
Download Presentation

DAA U-2(part1)

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Disjoint Sets and Dynamic Equivalence Dr.B.Srinu Dept of Computer Science & Engineering

  2. Today’s Outline • Disjoint Set Union/Find ADT • Up-trees • Weighted Union • Path Compression 2

  3. find(4) {1,4,8} {6} 8 {7} {2,3,6} {5,9,10} {2,3} union(2,6) Disjoint Set Union/Find ADT Union/Find ADT operations • union • find • create • destroy Disjoint set equivalence property: every element of a DS U/F structure belongs to exactly one set Dynamic equivalence property: the set of an element can change after execution of a union CSE 326 Autumn 2001 3

  4. Disjoint Set Union/FindMore Formally Given a set U = {a1, a2, … , an} • Maintain a partition of U, a set of subsets of U {S1, S2, … , Sk} such that: • each pair of subsets Si and Sj are disjoint: • together, the subsets cover U: • The Si are mutually exclusive and collectively exhaustive • Union(a, b) creates a new subset which is the union of a’s subset and b’s subset NOTE: outside agent decides when/what to union - ADT is just the bookkeeper • Find(a) returns a unique name for a’s subset NOTE: set names are arbitrary! We only care that: Find(a) == Find(b)  a and b are in the same subset CSE 326 Autumn 2001 4

  5. a b c 3 10 2 1 6 d e f 4 7 11 9 8 g h i 12 5 Example Construct the maze on the right Initial state (set names underlined): {a}{b}{c}{d}{e}{f}{g}{h}{i} Maze constructor (outside agent) traverses edges in random order and decides whether to union CSE 326 Autumn 2001 5

  6. 3 10 2 1 6 4 7 11 9 8 12 5 Example, First Step a b c {a}{b}{c}{d}{e}{f}{g}{h}{i} find(b) b find(e) e find(b)  find(e) so: add 1 to E union(b, e) {a}{b,e}{c}{d}{f}{g}{h}{i} d e f g h i Order of edges in blue CSE 326 Autumn 2001 6

  7. Up-Tree Intuition Finding the representative member of a set is somewhat like the opposite of finding whether a given item exists in a set. So, instead of using trees with pointers from each node to its children; let’s use trees with a pointer from each node to its parent. CSE 326 Autumn 2001 7

  8. Union-Find Up-Tree Data Structure • Each subset is an up-tree with its root as its representative member • All members of a given set are nodes in that set’s up-tree • Hash table maps input data to the node associated with that data a c h g f i d b e Up-trees are not necessarily binary! CSE 326 Autumn 2001 8

  9. Find find(f) find(e) a b c 10 a c g h d e f 7 f i d b 11 9 8 g h i 12 e Just traverse to the root! runtime: CSE 326 Autumn 2001 9

  10. Union union(a,c) a b c 10 a c g h d e f f i d b 11 9 8 g h i 12 e Just hang one root from the other! runtime: CSE 326 Autumn 2001 10

  11. a b c 3 10 Example (1/11) 2 1 6 d e f 4 7 11 9 8 union(b,e) g h i 12 5 a b c d e f g h i a b c d f g h i e

  12. a b c 3 10 Example (2/11) 2 6 d e f 4 7 11 9 8 union(a,d) g h i 12 5 a b c d f g h i e a b c f g h i d e

  13. a b c 3 10 Example (3/11) 6 d e f 4 7 11 9 8 union(a,b) g h i 12 5 a b c f g h i d e a c f g h i b d e

  14. a c f g h i b d e While we’re finding e, could we do anything else? a b c 10 Example (4/11) 6 d e f 4 7 11 9 8 find(d) = find(e) No union! g h i 12 5

  15. a b c 10 Example (5/11) 6 d e f 7 11 9 8 union(h,i) g h i 12 5 a c f g h i a c f g h b d i b d e e

  16. a b c 10 Example (6/11) 6 d e f 7 11 9 8 union(c,f) g h i 12 a c f g h a c g h b d b d f i i e e

  17. a c g h c g h b d f a f i i b d e Could we do a better job on this union? e a b c 10 Example (7/11) d e f 7 find(e) find(f) union(a,c) 11 9 8 g h i 12

  18. a b c 10 Example (8/11) d e f find(f) find(i) union(c,h) 11 9 8 g h i 12 c g h c g a f a f h i b d b d i e e

  19. a b c 10 Example (9/11) d e f find(e) = find(h) and find(b) = find(c) So, no unions for either of these. 11 9 g h i 12 c g a f h b d i e

  20. a b c Example (10/11) d e f find(d) find(g) union(c, g) 11 g h i 12 c g g c a f h a f h b d i b d i e e

  21. a b c Example (11/11) d e f find(g) = find(h) So, no union. And, we’re done! g h i 12 g a b c c d e f a f h g h i b d i Ooh… scary! Such a hard maze! e

  22. a c g h b d f i e 0 (a)1 (b)2 (c)3 (d)4 (e)5 (f)6 (g)7 (h)8 (i) -1 0 -1 0 1 2 -1 -1 7 Disjoint set data structure • A forest of up-trees can easily be stored in an array. • Also, if the node names are integers or characters, we can use a very simple, perfect hash. Nifty storage trick! up-index: CSE 326 Autumn 2001 22

  23. Implementation typedef ID int; ID find(Object x) { assert(hTable.contains(x)); ID xID = hTable[x]; while(up[xID] != -1) { xID = up[xID]; } return xID; } ID union(ID x, ID y) { assert(up[x] == -1); assert(up[y] == -1); up[y] = x; } runtime: O(depth) or … runtime: O(1) CSE 326 Autumn 2001 23

  24. a c g h b d f i e Room for Improvement:Weighted Union • Always makes the root of the larger tree the new root • Often cuts down on height of the new up-tree c g h a g h a f b d i c i b d f e Could we do a better job on this union? e Weighted union! CSE 326 Autumn 2001 24

  25. Weighted Union Code typedef ID int; ID union(ID x, ID y) { assert(up[x] == -1); assert(up[y] == -1); if (weight[x] > weight[y]) { up[y] = x; weight[x] += weight[y]; } else { up[x] = y; weight[y] += weight[x]; } } new runtime of union: new runtime of find: CSE 326 Autumn 2001 25

  26. Weighted Union Find Analysis • Finds with weighted union are O(max up-tree height) • But, an up-tree of height h with weighted union must have at least 2h nodes •  2max height n and max height  log n • So, find takes O(log n) Base case: h = 0, tree has 20 = 1 node Induction hypothesis: assume true for h < h A merge can only increase tree height by one over the smaller tree. So, a tree of height h-1 was merged with a larger tree to form the new tree. Each tree then has  2h-1 nodes by the induction hypotheses for a total of at least 2h nodes. QED. CSE 326 Autumn 2001 26

  27. a c f g h i b d e Room for Improvement: Path Compression • Points everything along the path of a find to the root • Reduces the height of the entire access path to 1 a c f g h i b d e While we’re finding e, could we do anything else? Path compression! CSE 326 Autumn 2001 27

  28. Path Compression Example find(e) c c g g a f h a f h b e b d d i i e CSE 326 Autumn 2001 28

  29. Path Compression Code typedef ID int; ID find(Object x) { assert(hTable.contains(x)); ID xID = hTable[x]; ID hold = xID; while(up[xID] != -1) { xID = up[xID]; } while(up[hold] != -1) { temp = up[hold]; up[hold] = xID; hold = temp; } return xID; } runtime: CSE 326 Autumn 2001 29

  30. Complexity of Weighted Union + Path Compression Ackermann created a function A(x, y) which grows very fast! Inverse Ackermann function (x, y) grows very slooooowwwly … Single-variable inverse Ackermann function is called log* n How fast does log n grow? log n = 4 for n = 16 Let log(k) n = log (log (log … (log n))) Then, let log* n = minimum k such that log(k) n  1 How fast does log* n grow? log* n = 4 for n = 65536 log* n = 5 for n = 265536 (20,000 digit number!) How fast does (x, y) grow? Even sloooowwwer than log* n (x, y) = 4 for n far larger than the number of atoms in the universe (2300) CSE 326 Autumn 2001 30

  31. Complexity of Weighted Union + Path Compression • Tarjan proved that m weighted union and find operations on a set of n elements have worst case complexity O(m(m, n)) • This is essentially amortized constant time • In some practical cases, weighted union or path compression or both are unnecessary because trees do not naturally get very deep. CSE 326 Autumn 2001 31

  32. SummaryDisjoint Set Union/Find ADT • Simple ADT, simple data structure, simple code • Complex complexity analysis, but extremely useful result: essentially, constant time! • Lots of potential applications • Object-property collections • Index partitions: e.g. parts inspection • To say nothing of maze construction • In some applications, it may make sense to have meaningful (non-arbitrary) set names CSE 326 Autumn 2001 32

More Related