Lunch will be served at 11:45 AM.
We address the problem of selling $k$ items to multiple unit-demand buyers to maximize revenue in a setting where the auctioneer can inspect the buyers' types. Our work diverges from classical mechanism design approaches, requiring a novel analysis. Using techniques from convex analysis and the calculus of variations, we fully characterize the optimal mechanism for a single buyer. Subsequently, in conjunction with prophet inequality results, we demonstrate how our analysis enables us to achieve $1-1/\sqrt{k+3}$ of the optimal revenue in expectation for the general problem. Based on paper https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3700525
Ali Makhdoumi is an Associate Professor at the Fuqua School of Business, Duke University. He holds a BSc in Electrical Engineering and a BSc in Mathematics from Sharif University of Technology, as well as a Ph.D. in Electrical Engineering and Computer Science from MIT. His research interests include optimization, game theory, stochastic modeling, and learning theory with applications to market design, mechanism design, privacy, and data markets.
LSRC D344 or join virtually via Zoom https://duke.zoom.us/j/92510717076