Inapproximability of Combinatorial Public Projects Michael Schapira (Yale University and UC Berkeley) Joint work with Yaron Singer (UC Berkeley) Overview of the Talk • The Combinatorial Public Project Problem. • The Submodular Case – Background. • A Trivial Truthful Approximation-Algorithm. • Our Main Result. • Conclusions and Open Questions. Algorithmic Mechanism Design • Algorithmic Mechanism Design deals with designing efficient mechanisms for decentralized computerized settings [Nisan-Ronen]. • Takes into account both the strategic behavior of the different participants and the usual computational efficiency considerations. • Target applications: protocols for Internet environments. Combinatorial Public Project • Set of n users; Set of m resources; • Each user i has a valuation function: vi : 2[m] → R≥0 • Objective: Given a parameter k, choose a set of resources S* of size k which maximizes the social welfare: S* = argmax Σi vi(S) S [m], |S|=k Assumptions Regarding Each Valuation Function • Normalized: v(∅ ) = 0 • Non-decreasing: v(S) ≤ v(T) S T • Subadditive: v( S) + v(T) ≥ v( S υ T) [ Submodular: v( S υ { j })− v(S) ≥ v( T υ { j })− v(T) S T ] Motivating Examples • Elections for a committee: The agents are voters, resources are potential candidates. • Overlay networks: We wish to select a subset of nodes in a graph that will function as an overlay network. [http://nms.csail.mit.edu/ron/] Access Models How can we access the input ? • One possibility: succinct valuations computational complexity approach. • The “black box” approach: each bidder is represented by an oracle which can answer certain queries. Communication complexity approach. What Do We Want? • Quality of the solution: As close to the optimum as possible. • Computationally tractable: Polynomial running time (in n and m). • Truthful: Motivate (via payments) agents to report their true values regardless of other agents’ reports. • The utility of each user is ui = vi(S) - pi The Submodular Case [Papadimitriou-S-Singer] • Computational Perspective: A 1-1/e approximation ratio is achievable due to the submodularity of the valuations (but not truthful) A tight lower bound exists [Feige]. • Strategic Perspective: A truthful solution is achievable via VCG payments (but NP-hard to obtain) • What about achieving both simultaneously? The Submodular Case: Truth and Computation Don’t Mix • Theorem [Papadimitriou-S-Singer]: Any truthful algorithm for the combinatorial public project problem which approximates better than √m requires exponential communication in m. Even for n=2. • Implications for AMD: A huge gap between truthful&polynomial algorithms, and truthful/polynomial algorithms. A trivial √m-approximation Algorithm for Subadditiver Agents • The algorithm: If k≤√m, simply choose the single resource j for which the social-welfare is maximized. If k>√m, divide the m resources to √m disjoint sets of equal size and choose the one that maximizes the social welfare. The Algorithm is Truthful • Fact: Maximal-in-range algorithms are truthful (VCG). • -> The trivial approximation algorithm is (essentially) the best truthful algorithm for the submodular case. Also for the subadditive case. Upper and Lower Bounds constant non- truthful upper bounds exist ? Submodular Subadditive √m truthful upper bound √m truthful upper bound High Hopes • Twin problem: combinatorial auctions. • Theorem (Informal): There is a 2approximation algorithm for combinatorial auctions with subadditive bidders. [Feige] Our Main Result • Theorem: Any approximation algorithm for the combinatorial public project problem with subadditive agents which approximates better than O(m1/4) requires exponential communication in m. • Implications: The trivial truthful approximation algorithm is nearly tight even from a purely computational perspective. Other Results and an Open Question • A communication complexity lower bound for general valuations. • A computational complexity lower bound for general valuations. • Open question: prove a computational-complexity analogue of our result. Thanks!