杨健教授系列讲座
报告人:杨健,2000 年于美国University of Texas, Austin 获管理学博士学位。现任职于罗格斯大学,商学院 管理科学与信息系统系,副教授。
#1 TITLE: A Link between Sequential Semi-anonymous Nonatomic Games and their Large but Finite Counterparts}
时间:2014年22日(周四)下午3:00 - 4:00
地点:数学楼二楼学术报告厅
ABSTRACT: We show that equilibria of a sequential semi-anonymous nonatomic game (SSNG) can be adopted by players in corresponding large but finite games to achieve near-equilibrium payoffs. Such equilibria are parsimonious in form and easy to execute, as they are both oblivious of past history and blind to other players' present states. To arrive at the result, we propose a concept of asymptotic resemblance to product measures and verify the resence of such resemblance in joint state-action distributions in large games. Then, when an SSNG with a finite horizon possesses an equilibrium in the form of a random state-to-action rule, we show that large but finite counterparts of the game can use the equilibrium to reach asymptotically equilibrium payoffs on average. The above transient result can be extended to a stationary case, where the finite counterparts are special discounted stochastic games. The kind of equilibria we adopt for SSNG are similar to distributional equilibria that are well understood in literature, and they themselves are shown to exist.
#2 TITLE: Joint Control of Emissions Permit Trading and Production with Fixed-cost Considerations
时间:2014年23日(周五)下午3:00 - 4:00
地点:数学楼二楼学术报告厅
ABSTRACT: The trading of emissions permits, as an effective market-based control of harmful emissions, is becoming widespread over the globe. A firm participating in emissions permit trading has only to balance its permits at the end of a planning horizon, say one year. We attempt to shed light on the optimal coordination of permit trading and permit-consuming production for a firm facing both Markov permit price
process and random demand during the multi-period planning horizon. %In each period, it can make trading and production decisions using information about
its on-hand permit and inventory levels as well as the market-prevalent permit price.
We focus on the case with non-negligible fixed transaction setup costs. Through the exploitation of $K$-convexity variants that are suitable for
two-dimensional control, we achieve partial characterization of optimal control policies. When the selling of permits is prohibited, we find that
$(s,S)$-type permit purchase behavior can be predicted. For the more general case involving two-way trading, we prescribe an optimal three-interval
trading policy. Heuristics, including one based on the uncoupling of trading and production activities, are introduced. %When it is the fixed cost for
production setup that becomes dominant, we identify targeting-chasing optimal trading policies and $(s,S)$-type production policies.
#3 TITLE: Inventory and Price Control under Time-consistent Coherent and Markov Risk Measure
时间:2014年27日(周二)下午3:00 - 4:00
地点:数学楼二楼学术报告厅
ABSTRACT: We use the recently proposed concept of time-consistent coherent and Markov risk measure on the study of a risk-averse firm's inventory and
price control activities. In our shock-driven setting which is different from the state-driven setting where the measure is first introduced, we show
the suitability of dynamic programming formulations. On this basis, we examine a joint inventory-price control problem. The resulting model calls
for worst-case analysis over a convex set of demand-distribution scenarios in every period. We achieve a structural characterization for an optimal
policy that is reminiscent of its risk-neutral counterpart. Monotone properties are derived for the pricing policy when the convex risk set %is
either in a product form or constitutes a lattice under a suitably defined partial order. We also introduce the concept of optimism using the strong
set order between risk sets. Two risk measures thus ranked produce inventory and pricing decisions that can be ranked themselves.