the power of two prices beyond cross monotonicity
play

The Power of Two Prices: Beyond Cross-Monotonicity - PowerPoint PPT Presentation

Cost-Sharing State of the Art The Power of Two Prices Conclusion The Power of Two Prices: Beyond Cross-Monotonicity Incentive-Compatible Mechanism Design Yvonne Bleischwitz Burkhard Monien Florian Schoppmann Karsten Tiemann Department of


  1. Cost-Sharing State of the Art The Power of Two Prices Conclusion The Power of Two Prices: Beyond Cross-Monotonicity Incentive-Compatible Mechanism Design Yvonne Bleischwitz Burkhard Monien Florian Schoppmann Karsten Tiemann Department of Computer Science University of Paderborn March 27, 2007 University of Paderborn Burkhard Monien Mar 27, 2007 1 / 24 · ·

  2. Cost-Sharing State of the Art The Power of Two Prices Conclusion The Model ◮ n ∈ N players: ◮ Have private valuations v i ∈ R ≥ 0 for service, v := ( v i ) i ∈ [ n ] ◮ Submit bids b i ∈ R ≥ 0 to service provider, b := ( b i ) i ∈ [ n ] ◮ Service provider uses mechanism to determine outcome: Definition (Cost-Sharing Mechanism) Mechanism Q ( b ) (“White Box”) b ( Q × x ) : player ≥ 0 → 2 [ n ] × R n x ( b ) R n set [ n ] ◮ Desirable that b = v but this cannot be a priori guaranteed University of Paderborn Burkhard Monien Mar 27, 2007 2 / 24 · ·

  3. Cost-Sharing State of the Art The Power of Two Prices Conclusion Common Assumptions for Cost-Sharing Mechanisms Only consider mechanisms with the following properties ∀ i ∈ [ n ] : ◮ NPT (No Positive Transfer) = no negative payments: x i ( b ) ≥ 0 ◮ VP (Voluntary Participation) = obey bids: x i ( b ) ≤ b i ◮ CS* (Strict Consumer Sovereignty): CS: ∃ b + i ∈ R ≥ 0 : ∀ b ∈ R n ≥ 0 : ( b i ≥ b + i = ⇒ i ∈ Q ( b )) Strictness: ∀ b ∈ R n ≥ 0 : ( b i = 0 = ⇒ i �∈ Q ( b )) Assume: v is true valuation vector, ( Q × x ) mechanism ◮ Player i ’s utility depends on bid vector: � v i − x i ( b ) if i ∈ Q ( b ) u i ( b ) := 0 if i / ∈ Q ( b ) University of Paderborn Burkhard Monien Mar 27, 2007 3 / 24 · ·

  4. Cost-Sharing State of the Art The Power of Two Prices Conclusion Desirable Properties of Cost-Sharing Mechanisms ◮ GSP (Group-Strategyproofness): ∀ true valuations v ∈ R n ≥ 0 : ∄ coalition K ⊆ [ n ] such that ∃ cheating possibility b K ∈ R K ≥ 0 with ◮ u i ( v − K , b K ) ≥ u i ( v ) for all i ∈ K and ◮ u i ( v − K , b K ) > u i ( v ) for at least one i ∈ K . SP: Needs to hold only for coalitions K of size 1 Definition ( n -Player Cost Function) Function C : 2 [ n ] → R ≥ 0 with C ( A ) = 0 ⇐ ⇒ A = ∅ ◮ β -BB ( β -Budget-Balance, with 0 ≤ β ≤ 1): � β · C ( Q ( b )) ≤ x i ( b ) ≤ OPT ( Q ( b )) i ∈ [ n ] University of Paderborn Burkhard Monien Mar 27, 2007 4 / 24 · ·

  5. Cost-Sharing State of the Art The Power of Two Prices Conclusion A Cost-Sharing Scenario Computing center with large cluster of parallel machines ◮ Offering customers (uninterrupted) processing times ◮ Cost proportional to makespan Machine A Customer 1 Machine B Customer 2 Customer 3 Machine C Customer 4 Customer 5 Makespan({1, 2, 3, 4, 5 }) = 6 University of Paderborn Burkhard Monien Mar 27, 2007 5 / 24 · ·

  6. Cost-Sharing State of the Art The Power of Two Prices Conclusion Implications of GSP GSP is a very strong requirement: ◮ Even coalitions with binding agreements should have no incentive to cheat Theorem (Moulin, 1999) Let ( Q × x ) be a GSP cost-sharing mechanism, b , b ′ ∈ R n ≥ 0 bid vectors with Q ( b ) = Q ( b ′ ) . Then x i ( b ) = x i ( b ′ ) for all i ∈ [ n ] . Hence, GSP (with standard assumptions NPT, VP, CS*) implies: ◮ Payments independent of bids ◮ Bids only determine set of serviced players University of Paderborn Burkhard Monien Mar 27, 2007 6 / 24 · ·

  7. Cost-Sharing State of the Art The Power of Two Prices Conclusion Cost-Sharing Methods Last theorem gives rise to: Definition ( n -Player Cost-Sharing Method) Function ξ : 2 [ n ] → R n ≥ 0 . ξ is cross-monotonic if ∀ A , B ⊆ [ n ] and ∀ i ∈ A : ξ i ( A ) ≥ ξ i ( A ∪ B ) Note: ◮ β -Budget-balance defined as before: � ∀ A ⊆ [ n ] : β · C ( A ) ≤ ξ i ( A ) ≤ OPT ( A ) i ∈ [ n ] University of Paderborn Burkhard Monien Mar 27, 2007 7 / 24 · ·

  8. Cost-Sharing State of the Art The Power of Two Prices Conclusion Moulin Mechanisms ≥ 0 → 2 [ n ] × R n (Moulin, 1999) Algorithm M ξ : R n Input: b ∈ R n ≥ 0 ; Output: Q ∈ 2 [ n ] , x ∈ R n 1: Q := [ n ] 2: while ∃ i ∈ Q : b i < ξ i ( Q ) do Q := { i ∈ Q | b i ≥ ξ i ( Q ) } 3: x := ξ ( Q ) Theorem (Moulin, 1999) M ξ satisfies GSP and β -BB if ξ is cross-monotonic and β -BB. University of Paderborn Burkhard Monien Mar 27, 2007 8 / 24 · ·

  9. Cost-Sharing State of the Art The Power of Two Prices Conclusion Submodular Cost Functions Definition (Submodular Cost-Function) Cost function C : 2 [ n ] → R ≥ 0 where for all A ⊆ B ⊆ [ n ] and i / ∈ B C ( A ∪ { i } ) − C ( A ) ≥ C ( B ∪ { i } ) − C ( B ) . Complete characterization when C submodular: Theorem (Moulin, 1999) Any GSP and 1-BB mechanism has cross-monotonic cost-shares. A 1-BB cross-monotonic ξ exists. Hence, M ξ is GSP and 1-BB. Submodular seems natural (“marginal costs only decrease”), but: ◮ Example: makespan scheduling C ([ 1 ]) = 1, C ([ 2 ]) = 1, C ([ 3 ]) = 1, C ([ 4 ]) = 2 University of Paderborn Burkhard Monien Mar 27, 2007 9 / 24 · ·

  10. Cost-Sharing State of the Art The Power of Two Prices Conclusion Previous Research Good BB. Examples for cross-monotonic cost-sharing methods: β − 1 Authors Problem Jain, Vazirani (2001) MST 1 Steiner tree, TSP 2 Pál, Tardos (2003) Facility location 3 Single-Source-Rent-or-Buy 15 Gupta et. al. (2003) Single-Source-Rent-or-Buy 4 . 6 Könemann et. al. (2005) Steiner forest 2 2 m Bleischwitz, Monien (2006) Scheduling on m links m + 1 . . . University of Paderborn Burkhard Monien Mar 27, 2007 10 / 24 · ·

  11. Cost-Sharing State of the Art The Power of Two Prices Conclusion A Note on Modeling Assumptions Recall: ◮ CS: ∃ b + ≥ 0 : ( b i ≥ b + i ∈ R ≥ 0 : ∀ b ∈ R n i = ⇒ i ∈ Q ( b )) ◮ CS*: CS and also ∀ b ∈ R n ⇒ i �∈ Q ( b )) ≥ 0 : ( b i = 0 = Trivial GSP, 1-BB mechanism if only CS (Immorlica et. al., 2005): ◮ “Taking a fixed order, find 1 st agent who can pay for the rest” Even stronger than CS*: ◮ NFR (No Free Riders): i ∈ Q ( b ) = ⇒ x i ( b ) > 0 University of Paderborn Burkhard Monien Mar 27, 2007 11 / 24 · ·

  12. Cost-Sharing State of the Art The Power of Two Prices Conclusion Symmetric Costs With CS*, it is much harder to achieve GSP and good BB. Does symmetry of costs help? That is, for A , B ⊆ [ n ] we have | A | = | B | = ⇒ C ( A ) = C ( B ) . We define c : [ n ] → R ≥ 0 , c ( i ) := C ([ i ]) in this case. Our results (not discussed in this talk): ◮ We give a general GSP, 1-BB mechanism for 3 or less players ◮ There is a 4-player symmetric cost function for which no GSP, 1-BB mechanism exists University of Paderborn Burkhard Monien Mar 27, 2007 12 / 24 · ·

  13. Cost-Sharing State of the Art The Power of Two Prices Conclusion The Power of Two Prices Bleischwitz, Monien (2006): For makespan costs (weights or machines identical), cross-monotonic methods are no better than m + 1 2 m -BB in general ◮ Is there a mechanism that is better than Moulin here? (Recall: Makespan is not submodular function) ◮ Is it a generic mechanism? Yes, if the cost function is symmetric. University of Paderborn Burkhard Monien Mar 27, 2007 13 / 24 · ·

  14. Cost-Sharing State of the Art The Power of Two Prices Conclusion Cost-Sharing Forms (1/2) ◮ Preference order. Cost vectors ξ j ∈ R j ≥ 0 , 1 2 3 4 5 j ∈ [ n ] , such that for i ∈ [ n ] , A ⊆ [ n ] : ξ | A | = (4, 2, 2) � ξ | A | if i ∈ A Rank ( i , A ) ξ i ( A ) := 1 2 3 4 5 0 otherwise. ◮ At most 2 different cost-shares for any set of players A ⊆ [ n ] Definition (Cost-Sharing Form) Consists of: Sequence ( a k , λ k ) k ∈ N ⊂ R 2 > 0 , mappings σ : N → N , f : N → N 0 A cost-sharing form defines cost vectors ξ i , i ∈ N : ξ i = ( λ σ ( i ) , . . . , λ σ ( i ) , a σ ( i ) , . . . , a σ ( i ) ) � �� � f ( i ) elements University of Paderborn Burkhard Monien Mar 27, 2007 14 / 24 · ·

  15. Cost-Sharing State of the Art The Power of Two Prices Conclusion Cost-Sharing Forms (2/2) Recall: A cost-sharing form defines cost vectors ξ i , i ∈ N : ξ i = ( λ σ ( i ) , . . . , λ σ ( i ) , a σ ( i ) , . . . , a σ ( i ) ) � �� � f ( i ) elements Valid cost-sharing form: Example: ◮ σ ( i + 1 ) ∈ { σ ( i ) , σ ( i ) + 1 } ξ i i f ( i ) σ ( i ) ◮ σ ( i + 1 ) = σ ( i ) + 1 1 0 1 ( 2 ) = ⇒ f ( i + 1 ) = 0 2 0 1 ( 2 , 2 ) 3 1 1 ( 3 , 2 , 2 ) ◮ f ( 1 ) = 0 4 2 1 ( 3 , 3 , 2 , 2 ) ◮ f ( i + 1 ) ≤ f ( i ) + 1 5 0 2 ( 1 , 1 , 1 , 1 , 1 ) ◮ λ k ≥ a k ≥ a k − 1 6 1 2 ( 5 , 1 , 1 , 1 , 1 , 1 ) σ induces segments: Ranges of cardinalities with same cost-shares! University of Paderborn Burkhard Monien Mar 27, 2007 15 / 24 · ·

Download Presentation
Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.

Recommend


More recommend