1 / 79

Cake Cutting is and is not a Piece of Cake

Cake Cutting is and is not a Piece of Cake. Jeff Edmonds , York University Kirk Pruhs, University of Pittsburgh. Toronto Star. Informal Problem Statement. Resource allocation between n self possibly deceitful players. I like. I like. 0. 1. Classic Problem Definition. = [0, 1].

eliza
Download Presentation

Cake Cutting is and is not a Piece of Cake

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. Cake Cutting is and is not a Piece of Cake Jeff Edmonds, York University Kirk Pruhs, University of Pittsburgh

  2. Toronto Star

  3. Informal Problem Statement Resource allocation between n self possibly deceitful players

  4. I like I like 0 1 Classic Problem Definition = [0, 1] • n players wish to divide a cake • Each player p has an unknown value function Vp

  5. v I like x y 0 1 Classic Problem Definition = [0, 1] • n players wish to divide a cake • Each player p has an unknown value function Vp • Allowed Operations : • Eval[p, x, y]: returns how much player p values piece/interval [x, y]

  6. v I like x y 0 1 Classic Problem Definition = [0, 1] • n players wish to divide a cake • Each player p has an unknown value function Vp • Allowed Operations : • Eval[p, x, y]: returns how much player p values piece/interval [x, y] • Cut[p, x, v]: returns a y such Eval[p,x, y] = v

  7. 1/n I like 0 1 Classic Problem Definition = [0, 1] • n players wish to divide a cake • Each player p has an unknown value function Vp • Goal: Fair cut Each honest player p is guaranteed a piece of value at least 1/n.

  8. 1/n I like 0 1 Classic Problem Definition = [0, 1] • n players wish to divide a cake • Each player p has an unknown value function Vp • Goal: Fair cut Each honest player p is guaranteed a piece of value at least 1/n.

  9. History • Originated in 1940’s school of Polish mathematics • Picked up by social scientists interested in fair allocation of resources • The subject of a handful of books (e.g Brams and Taylor, and Robertson and Webb) and many papers • A quick Google search reveals cake cutting is used as a teaching example in many algorithms courses, e.g CMU CS 15-451 and CMU CS 15-750

  10. I like I like I’ll take the right. Classic Alg: Cut and Choose (n=2) • Person A cuts the cake into two pieces • Person B selects one of the two pieces, and person A gets the other piece Careful. My half cut is here. I am a little jealous but I am happywith the left.

  11. I like I like O(n log n) Divide and Conquer Algorithm: Evan and Paz • Yp = cut(p, 0, 1/2) for p = 1 … n My half cut is here. My half cut is here.

  12. I like O(n log n) Divide and Conquer Algorithm: Evan and Paz • Yp = cut(p, 0, 1/2) for p = 1 … n My half cut is here.

  13. O(n log n) Divide and Conquer Algorithm: Evan and Paz • Yp = cut(p, 0, 1/2) for p = 1 … n • m = median(y1, … , yn)

  14. I like I like so am happy with the left. so am happy with the right. O(n log n) Divide and Conquer Algorithm: Evan and Paz • Yp = cut(p, 0, 1/2) for p = 1 … n • m = median(y1, … , yn) • Recurse on [0, m] with those n/2 players p for which yp < m • Recurse on [m, 1] with those n/2 players p for which yp > m • Time O(nlogn)

  15. Problem Variations • Contiguousness: Assigned pieces must be subintervals • Approximate fairness: A protocol is 1/c-fair if each player is a assured a piece that he gives a value of at least 1/cn • Approximate queries (introduced by us?): • AEval[p, ε, x, y]: returns a value v such that Vp[x, y]/(1+ε) ≤ v ≤ (1+ ε) Vp[x, y] • ACut[p, ε, x, v]: returns a y such Vp[x, y]/(1+ε) ≤ v ≤ (1+ ε) Vp[x, y]

  16. Problem Variations (Approximate) • Open: • Remove contiguous requirement for randomized alg, • Algorithm’s failure prob (1) vs 1/n(1) My grad student Jai.

  17. Outline • Deterministic Ω(n log n) Lower Bound • Definition of Thin-Rich Problem • Ω(log n) lower bound for Thin-Rich • (Skip) Randomized with Approximate Cuts Ω(n log n) Lower Bound • Randomized with Exact Cuts O(n) Upper Bound

  18. At least n/2 players require thin rich piece Thin-Rich Game • Game Definition: Single player must find a thin rich piece. • A piece is thin if it has width ≤ 3/n • A piece is rich if it has value ≥ 1/3n • Theorem: The deterministic complexity of Thin-Rich is Ω(log n). • Theorem: The deterministic complexity of cake cutting is Ω(nlog n).

  19. Alg Adv • I dynamically choose how to answer • and the value function • so after O(logn) operationsAlg cant find a thin rich piece. • I give sequence of Eval[x,y] & Cut[x,v]operations.

  20. Alg Adv (log n) ... ... ... ... ... ... • I build a complete 3-ary treewith the n pieces as leaves

  21. Alg Adv ½ ¼ ½ ¼ ¼ ¼ (log n) ... ... ... ... ... ... • For each node, • I label edges • <½,¼,¼> or <¼,¼,½>

  22. Alg Adv ½ ¼ ½ ¼ ¼ ¼ ¼ ¼ ½ (log n) ... ... ¼ ¼ ... ... ... ½ ... 1/1024 = ¼×¼×½×¼×¼×½ • Value of each piece isproduct of edge labelsin path.

  23. Alg Adv ½ ¼ ½ ¼ ¼ ½ ¼ ¼ ½ (log n) ... ... ¼ ¼ ... ... ... ½ ... 1/256 = ½×¼×½×¼×¼×½ To get a rich pieceI need at least 40% of the edge labels in path be ½. Good luck

  24. Alg Adv 0.4398 0 y • I need to find a yso that V[0,y] = 0.4398. • Cut[0,0.4398]?

  25. Alg Adv 1/2 0 1 0.4398 0.4398 • I do binary search to find0.4398 • Cut[0,0.4398]? 1/4

  26. Alg Adv ¼ ½ ¼ 1/4 2/4 0.4398 • I do binary search to find0.4398 • I fix some edge labels • Cut[0,0.4398]?

  27. Alg Adv ¼ ½ ¼ 0.4398 0.4398 • I do binary search to find0.4398 • I fix some edge labels • Cut[0,0.4398]? 1/4 2/4 6/16 7/16 4/16 8/16

  28. Alg Adv ¼ ½ ½ ¼ ¼ ¼ 7/16 8/16 0.4398 • I do binary search to find0.4398 • I fix some edge labels • Cut[0,0.4398]?

  29. Alg Adv ¼ ¼ ½ ½ ½ ¼ ¼ ¼ ¼ 0.4398 29/64 28/64 • I do binary search to find0.4398 • I fix some edge labels • Cut[0,0.4398]?

  30. Alg Adv ¼ ¼ ¼ ½ ½ ½ ½ ¼ ¼ ¼ ¼ ¼ 0.4398 112/256 113/256 • I do binary search to find0.4398 • I fix some edge labels • Cut[0,0.4398]?

  31. Alg Adv ¼ ¼ ¼ ½ ½ ½ ½ ½ ¼ ¼ ¼ ¼ ¼ ¼ ¼ 0.4398 450/1024 451/1024 • I do binary search to find0.4398 • I fix some edge labels • Cut[0,0.4398]?

  32. Alg Adv ¼ ¼ ¼ ½ ½ ½ ½ ½ ¼ ¼ ¼ ¼ ¼ ¼ ¼ 0.4398 1800/4096 1801/4096 ¼ ¼ ½ • I do binary search to find0.4398 • I fix some edge labels • Cut[0,0.4398]?

  33. Alg Adv ¼ ¼ ¼ ½ ½ ½ ½ ½ ¼ ¼ ¼ ¼ ¼ ¼ ¼ 0.4398 y ¼ ¼ ½ • I find a yso that V[0,y] = 0.4398. • Cut[0,0.4398]?

  34. Alg Adv ¼ ¼ ¼ ½ ½ ½ ½ ½ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ½ • I learned a path,but all its labels are ¼ • Hee Hee

  35. Alg Adv ¼ ¼ ¼ ½ ½ ½ ½ ½ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ½ • YesAfter t operationsevery path has only t known ½ labels. • Every path has  oneknown ½ label.

  36. Alg Adv ¼ ¼ ¼ ½ ½ ½ ½ ½ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ½ y x • I fix labels in path to x & y to ¼. • and give • Eval[x,y] = 0.00928 • Eval[x,y] 0.00928

  37. Alg Adv ¼ ¼ ¼ ½ ½ ½ ½ ½ ¼ ¼ ¼ ¼ ¼ ¼ ¼ (log n) ¼ ¼ ¼ ¼ ½ ¼ ¼ ¼ ¼ ½ That is bad.To get a rich pieceI need at least 40% of the edge labels in path be ½. • YesAfter t operationsevery path has only t½ known labels.

  38. At least n/2 players require thin rich piece Deterministic Ω(nlog n) Lower Bound • Theorem: The deterministic complexity of Thin-Rich is Ω(log n). • Theorem: The deterministic complexity of cake cutting is Ω(n log n).

  39. Outline • Deterministic Ω(n log n) Lower Bound • Randomized with O(n) Upper Bound • As intuition for the power of randomness, given a randomized algorithm for Thin-Rich with O(1) complexity • Describe “Balls and Bins” special case of cake cutting • Give “random graph” proof of “Balls and Bins” case • Give the flavor of the “random graph” proof for Cake Cutting

  40. O(1) Complexity Randomized Protocol for Thin-Rich • Pick an i uniformly at random from 0 … 3n-1 • x = Cut[0,i/3n] • y = Cut[ 0, (i+1)/3n] • Return [x,y] value = 1/3n (ie rich) Exp(width) = 1/3n (ie thin) 3n Candidate Pieces i x=i/3n y=(i+1)/3n

  41. Randomized Protocol for Cake Cutting Protocol Description: • Each player randomly selects 8 candidate pieces. • For each player, we carefully pick one of these

  42. Randomized Protocol for Cake Cutting Protocol Description: • Each player randomly selects 8 candidate pieces. • For each player, we carefully pick one of these • so that every point of cake is covered by at most O(1) pieces. • Where there is overlap, recurs. Works with probability Ω(1).

  43. Balls and Bins Two Random Choices: n balls, 3n bins each ball randomly chooses two bins. # Hits: Whp no bin has more than O(logn) balls. Choice: Select one of two bins for each ball. Online: Whp no bin has more than O(loglogn) balls. Offline: Whp no bin has more than O(1) balls.

  44. Proof on Offline: O(1)-Balls • Consider a graph G • Vertices = bins • One edge for each ball connecting the corresponding vertices • Important: Edges are independent • Lemma: If G is acyclic then the maximum load is 1

  45. Proof on Offline: O(1)-Balls • Classic Theorem: If a graph G has n/3 independent edges, then G is acyclic with (1) prob. • Proof: Prob[G contains a cycle C] ≤ ΣC Prob[C is in the graph] ~ Σi (n choose i) * (1/3n)i=(1).

  46. Cake Protocol  Ball & Bin Game Same if • Each player has uniform value on cake, so all candidate pieces are the same. • Each player randomly selects 2 (or 8) candidate pieces. • For each player, pick one of these. • O(1) over lap = O(1) balls per bin

  47. I like I like I like Cake Protocol  Ball & Bin Game Different if • Different cake values can make strangely overlapping candidate pieces.

  48. Cake Protocol  Ball & Bin Game Different if • Different cake values can make strangely overlapping candidate pieces.

  49. Vertex Vertex Vertex Vertex Directed Graph for Cake Cutting (choose 2 not 8 candidate peices) If take Must take Can’t take

  50. Must take If take Must take Must take One Difficulty Theorem: Sufficient Condition • If there is not directed path between the two pieces of the same person • then maximum load is at most 1

More Related