Jason R. Marden's Webpage

Overview of Topics

In this paper we break papers out by different topic areas.  Please click here If you prefer to see a list of papers in chronological order.  The main areas area as follows:

Chronological Order: Here is a complete list of papers coming out of the lab broken down by year of publication.

Book Chapters and Magazine Articles: This papers give a broad background on the role of game theory in distributed control of multiagent systems.

Load Shifting for Industrical Refrigeration: An emerging focus of my research centers on control theoretic advances for industrical refrigeration. While our literature in this area is relatively small and new, more will be added shortly. Plus, our team (UCSB, CrossnoKaye, and Lineage Solutions) received a 3 year grant for the California Energy Commission for this problem.  

Decision-Making in Complex Adversarial Environments: This literature provides a recent research thrust in decision-making in adversarial enviroments. Specific interest focuses on characterizing the "value of information" in such settings.  The primary domain considered here is Colonel Blotto Games, which are experiencing a resurgence in research attention in the past 5-10 years.  

Mechanism Design for Multiagent Systems:  Mechanism design, or more generally utility design, involves the design of utility functions for the agents in a system of interest. Traditionally, this field has focuses on auction theory where the auctioneer defines the auction, which in turn defines the utility function. However, this field is far broader as utility design is applicable to cooperative control (e.g., the assignment of local objective functions to programmable entities in a multi-agent system) and incentive design in societal bases systems (e.g., taxes in transportation networks).

Robust Social Infuence: This stream of work focuses on the derivation of incentive mechanisms, e.g., taxes or informational broadcasts, geared at influencing self-interested behavior and improving the emergent collective behavior.  Specific interest here is dedicated to robust incentive mechanisms, i.e., incentive mechanisms that are to be implement in systems where there are uncertainties in how self-interested users will respond to such mechanisms.  

The Role of Information in Multiagent Systems: This recent stream of work focuses on identifying the value of information in multiagent systems.  Multi-agent systems operate in environments where individual agents have access to locally avaialble information. The "degree" of this information clearly impacts that potential efficiency guarantees associated with agreeable networked control algorithms that rely on this information.  Accordingly, here we seek to identify how information imposes constraints on achievable performance and what information should be shared amongst the agents to improve these guarantees.     

Learning in Games: This stream of work focuses on the derivation of multi-agent decision making rules that provide guarantees on the resulting asymptotic behaviors.  Specific focus is on equilibrium selection, which involves driving the collective behavior to certain classes of equilibria that maybe more desirable from a societal perspective.

Cooperative Control of Networked Multiagent Systems: This stream of literature focuses on the design of networked control algorithms that ensure the agents resulting collective behavior is efficient with regards to a given system-level objective function. Applications of cooperative control will also be contained within.

Theses: This is where it all began!