Computing market equilibria is a problem of both theoretical and applied interest. Much research focuses on the static case, but in many markets items arrive sequentially and stochastically. We focus on the case of online Fisher markets: individuals have linear, additive utility and items drawn from a distribution arrive one at a time in an online setting. We define the notion of an equilibrium in such a market and provide a dynamics which converges to these equilibria asymptotically. An important use-case of market equilibria is the problem of fair division. With this in mind, we show that our dynamics can also be used as an online item-allocation rule such that the time-averaged allocations and utilities converge to those of a corresponding static Fisher market. This implies that other good properties of market equilibrium-based fair division such as no envy, Pareto optimality, and the proportional share guarantee are also attained in the online setting. An attractive part of the proposed dynamics is that the market designer does not need to know the underlying distribution from which items are drawn. We show that these convergences happen at a rate of $O(tfrac{log t}{t})$ or $O(tfrac{(log t)^2}{t})$ in theory and quickly in real datasets.
Download