Faculty Recruiting Support CICS

Online Linear Programming with Inventory Management with Application to Energy Procurement in Data Centers

06 Mar
Wednesday, 03/06/2019 12:15pm to 1:15pm
Computer Science Building Room 140
Theory Seminar

This talk considers the problem of online linear programming with inventory management constraints. Specifically, a decision maker should satisfy some units of an asset as her demand, either form a market with time-varying price or from her own inventory. The decision maker has presented a price in a slot-by-slot manner, and must immediately decide the purchased amount with the current price to cover the demand or to store in inventory for covering the future demand. The inventory has a limited capacity and its critical role is to buy and store assets at low price and use the stored assets to cover the demand at high price. The ultimate goal of the decision maker is to cover the demands while minimizing the cost of buying assets from the market. We propose several online algorithms for different scenarios and inventory models. We then show that the proposed algorithms achieve optimal competitive ratios, meaning that no other online algorithm can achieve a better theoretical guarantee. The underlying theoretical problems are motivated by the state-of-the-art applications of designing energy procurement strategies for data centers.

Faculty Host