1 / 24

Communication requirements of VCG-like mechanisms in convex environments

Communication requirements of VCG-like mechanisms in convex environments. Ramesh Johari Stanford University Joint work with John N. Tsitsiklis, MIT. Motivation. Resource allocation mechanisms with scalar strategy spaces: -single price: eff. loss · 25% (J & Tsitsiklis)

feng
Download Presentation

Communication requirements of VCG-like mechanisms in convex environments

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. Communication requirements of VCG-like mechanisms in convex environments Ramesh Johari Stanford University Joint work with John N. Tsitsiklis, MIT

  2. Motivation Resource allocation mechanismswith scalar strategy spaces: -single price: eff. loss · 25% (J & Tsitsiklis) -price differentiation: no eff. loss (Yang & Hajek, Maheswaran & Basar) This talk: generalization ofthe price differentiation case

  3. Outline • Resource allocation model • VCG mechanisms • Scalar strategy VCG mechanisms • Multicommodity flow models • Extensions and related work

  4. I Resource allocation model • Nusers • Jresources • Feasible allocations: X= { x2RN : x¸ 0, gj(x) · 0, j = 1, …, J } • gj(¢) : convex, differentiable • Assume: Slater condition holds

  5. I Utilities and payoffs • User r : utility Ur(xr) from allocation xr • concave, strictly increasing, differentiable • Payment to user r : tr • User r’s payoff (in $$$): Pr(xr, tr) = Ur(xr) + tr )Efficient allocation:

  6. II Achieving efficiency • In general: utilities are unknown • Design payments tr to alignefficiency and incentives • Planner wants to maximize: • User r wants to maximize:

  7. II VCG mechanisms • Strategy of user r:declared utilityVr • Mechanism chooses x(V) s.t.: • Payment to user r:

  8. II VCG mechanisms • Strategy of user r:declared utilityVr • Mechanism chooses x(V) s.t.: • Payment to user r:

  9. II VCG mechanisms • Strategy of user r:declared utilityVr • Mechanism chooses x(V) s.t.: • User r chooses Vr to maximize:

  10. II VCG mechanisms • Moral:truthful declaration is adominant strategy • Problem: Strategy spaces are overly complex • Main insight (for Nash implementation): Suffices to elicit only local derivativeof utility function

  11. III SSVCG mechanisms VCG-like with scalar strategy spaces. Parameterized family U(x ; ) s.t.: • xU(x ; ) is strictly concave • and strictly increasing, continuous, differentiable • “Slope matching”: 8 > 0 and x¸ 0, 9 > 0 s.t. U’(x; ) = 

  12. III SSVCG mechanisms • Mechanism chooses x() s.t.: • Payment to user r:

  13. III SSVCG: Key lemma is a Nash equilibriumif and only if for all r: Proof idea: If x* is optimal, user r can choose r s.t. Ur’(xr*) = U’(xr*; r)

  14. III SSVCG: Key lemma is a Nash equilibriumif and only if for all r: Proof idea: If x* is optimal, user r can choose r s.t. Ur’(xr*) = U’(xr*; r) tr

  15. III SSVCG: Efficient NE Corollary: Efficient Nash equilibrium exists Proof idea: Given efficient x*,each user r chooses r s.t. Ur’(xr*) = U’(xr* ; r) )Local truthful declaration

  16. III SSVCG: Efficient NE But: all NE are not efficient! Example: Single resource of capacity C = 1 User 1 bids hugeU’(C ; 1) All other users: Best response is to “give up” ) User 1 gets everything,regardless of true utilities

  17. III SSVCG: Efficient NE Given: NE  Define: P = { s : xs() > 0 } J = { j : gj(x()) = 0 } d(r) = ( gj/xr, j2J ) Theorem:If for all r, d(r) is linearly dependent on d(s), sr, s2P,then x() is efficient

  18. III SSVCG: Efficient NE We know: x() = maxx2XrU(xr ; r) For all r: x() 2 maxx2XUr(xr) + srU(xs ; s) First order conditions + linear dependence assumption ) Ur’(xr()) = U(xr() ; r)

  19. IV Networks Jlinks Capacity of link j : Cj User r$ subset of links X = { x¸ 0 : r : j2rxr·Cj, for all j } Assume: For all j, two users r1(j), r2(j), s.t. Uri(j)’(0) = 1 and r1(j) = r2(j) = {j} Then all NE allocations are efficient

  20. V Extensions • If Ur depends on k-dimensional xr : Need k-dimensional r • Designing hr(-r) is similar to VCG: Budget balance, etc.

  21. V Related work Yang & Hajek (2004),Maheswaran & Basar (2004): • Single resource, capacity = 1 • User r chooses bid r • Allocation: xr() = r / ss • User r pays: tr() = -r(ss) • Same as: SSVCG where U(x; ) = - (/x)

  22. V Related work Reichelstein and Reiter (1988): • more general environments: not quasilinear,no “aggregated goods” • mechanism is asymmetric: one user treated differently than others • requires (J - 1) + J/(N(N-1)) dimensional strategy space per user

  23. V Related work • Semret (1999) • Groves and Ledyard (1979) • Yang and Hajek (2005) • independent discovery of similar result

  24. III SSVCG: Efficient NE But: all NE are not efficient! Example: Two users, U1’(1) = 2, U2’(1) = 3 X = { (x1, x2) : 5x1+x2·6, x1+x2·1 } Then: • (1, 1) is not an efficient allocation • (1, 1) is an NE allocation:  s.t. U’(1 ; 1) = 4 ; U’(1 ; 2) = 1

More Related