1. Combinatorial Auction

2. Conbinatorial auction t 1 =20 t 2 =15 t 3 =6 f(t): the set X  F with the highest total value the mechanism decides the set of k winners and the corresponding payments Each player wants a bundle of objects t i : value player i is willing to pay for its bundle if player i gets the bundle at price p his utility is u i =t i -p F={ X  {1,…,N} : winners in X are compatible} v 1 =20 v 2 =16 v 3 =7

3. Combinatorial Auction (CA) problem – single-minded case Input: n buyers, m indivisible objects each buyer i: Wants a subset S i of the objects has a value t i for S i Solution: X  {1,…,n}, such that for every i,j  X, with i  j, S i  S j =  Measure (to maximize): Total value of X:  i  X t i

4. CA game each buyer i is selfish Only buyer i knows t i (while S i is public) We want to compute a “good” solution w.r.t. the true values We do it by designing a mechanism Our mechanism: Asks each buyer to report its value v i Computes a solution using an output algorithm g( ٠ ) takes payments p i from buyer i using some payment function p

5. More formally Type of agent buyer i: t i : value of S i Intuition: t i is the maximum value buyer i is willing to pay for S i Buyer i’s valuation of X  F: v i (t i,X)= t i if i  X, 0 otherwise SCF: a good allocation of the objects w.r.t. the true values

6. How to design a truthful mechanism for the problem? Notice that: the (true) total value of a feasible X is:  i v i (t i,X) the problem is utilitarian! …VCG mechanisms apply

7. VCG mechanism M= : g(r): arg max x  F  j v j (r j,x) p i (x): for each i: p i =  j≠i v j (r j,g(r - i )) -  j≠i v j (r j,x) g(r) has to compute an optimal solution… …can we do that?

8. Approximating CA problem within a factor better than m 1/2-  is NP-hard, for any fixed  >0. Theorem proof Reduction from independent set problem

9. Maximum Independent Set (IS) problem Input: a graph G=(V,E) Solution: U  V, such that no two verteces in U are jointed by an edge Measure: Cardinality of U Approximating IS problem within a factor better than n 1-  is NP-hard, for any fixed  >0. Theorem

10. the reduction CA instance has a solution of total value  k if and only if there is an IS of size  k G=(V,E) each edge is an object each node i is a buyer with: S i : set of edges incident to i t i =1 …since m  n 2 …

11. How to design a truthful mechanism for the problem? Notice that: the (true) total value of a feasible X is:  i v i (t i,X) the problem is utilitarian! …but a VCG mechanism is not computable in polynomial time! what can we do? …fortunately, our problem is one parameter!

12. A problem is binary demand (BD) if 1. a i ‘s type is a single parameter t i  2. a i ‘s valuation is of the form: v i (t i,o)= t i w i (o), w i (o)  {0,1} work load for a i in o when w i (o)=1 we’ll say that a i is selected in o

13. An algorithm g() for a maximization BD problem is monotone if  agent a i, and for every r -i =(r 1,…,r i-1,r i+1,…,r N ), w i (g(r -i,r i )) is of the form: Definition 1 Ө i (r -i ) riri Ө i (r -i )  {+  }: threshold payment from a i is: p i (r)= Ө i (r -i )

14. Our goal: to design a mechanism satisfying: 1. g( ٠ ) is monotone 2. Solution returned by g( ٠ ) is a “good” solution, i.e. an approximated solution 3. g( ٠ ) and p( ٠ ) computable in polynomial time

15. A greedy  m-approximation algorithm 1. reorder (and rename) the bids such that 2. W   ; X   3. for i=1 to n do 1. if S i  X=  then W  W  {i}; X  X  {S i } 4. return W v 1 /  |S 1 |  v 2 /  |S 2 |  …  v n /  |S n |

16. The algorithm g( ) is monotone Lemma proof It suffices to prove that, for any selected agent i, we have that i is still selected when it raises its bid Increasing v i can only move bidder i up in the greedy order, making it easier to win

17. How much can bidder i decrease its bid before being non- selected? Computing the payments …we have to compute for each selected bidder i its threshold value

18. Computing payment p i v 1 /  |S 1 |  …  v i /  |S i |  …  v n /  |S n | Consider the greedy order without i index j Use the greedy algorithm to find the smallest index j such that: 1. j is selected 2. S j  S i  p i = v j  |S i |/  |S j |

19. Let OPT be an optimal solution for CA problem, and let W be the solution computed by the algorithm, then Lemma iWiW  i  OPT v i   m  i  W v i proof OPT i ={j  OPT : j  i and S j  S i  }  i  W OPT i =OPT since it suffices to prove:  j  OPT i v j   m v i crucial observation for greedy order we have v i  |S j | iWiW  j  OPT i  |S i | v j 

20. proof we can bound Cauchy–Schwarz inequality iWiW  j  OPT i v j   j  OPT i vivi  |S i |  |S j |  j  OPT i  |S j |   |OPT i |  j  OPT i |S j | ≤|S i | ≤ m   m v i   |S i |  m

21. Cauchy–Schwarz inequality y j =  |S j | x j =1 n= |OPT i | for j=1,…,|OPT i | …in our case…