Meeting Minutes 20100727

=Meeting Minutes=
 * Meeting: TAC Energy Introduction
 * Date: July 27, 2010, 11:30pm CST
 * Participants: John Collins, Wolf Ketter, Prashant Reddy

Introductions (15 min)

 * Welcome from Wolf.
 * Prashant explains his background and interest in TAC Energy (background info at http://www.cs.cmu.edu/~ppr).
 * Prashant conveys interest in TAC Energy from his PhD advisor, Manuela Veloso (http://www.cs.cmu.edu/~mmv).
 * John presents an overview of TAC Energy with a specific focus on the Market Intelligence Service (MIS) and Negotiations.

Discussion (40 min)

 * John and Wolf present the current status of TAC Energy in terms of architectural modules.
 * John and Wolf present the intended timeline -- code complete by December for competition in 2011 summer.
 * John explains the decision to integrate Repast as a means of enabling non-computer scientists to participate in the competition.
 * John and Wolf highlight the need to conceptualize and model customer preferences/profiles/behaviors as a key aspect of the MIS.
 * Discussion on the dual role of the simulation infrastructure as a basis for the competition but also as a platform for ongoing research (e.g. custom customer models).
 * Discussion on the lessons learnt from TAC SCM and how those lessons were incorporated into the design of TAC Energy.
 * Brief discussion on TAC SCM's constraint of 6 participants -- need to allow for variable number of brokers in TAC Energy.


 * Brief discussion on mixed initiative:
 * Purely server-initiated communication versus Server- and Agent-initiated communication
 * Automated agent participation only versus human participation through interactive UIs.
 * Brief discussion of deployment architecture:


 * Latency issues that arise from each participant connecting from their servers
 * Where PDA might be a more practical market design than CDA because of the latency issues.

Next Steps (20 min)

 * Overview of current contributors:
 * Carsten Block, David Dauer and Daniel Schurr at KIT: Initial core framework.
 * John Collins (+ students) at Minnesota: High level architecture for broker agent, integration of Repast, and protocols for RFQ/Negotiations.
 * Andreas Symeonidis (+ team): Integrate Repast with server framework.
 * Wolf (+ team):
 * Sebastian: Master's thesis on modelling individual behaviors (esp. shifting) for households.
 * Yixin Lu: Ongoing PhD work in preference modelling.
 * Arun Batchu: Information architecture and integration architecture.
 * +1 software developer.


 * Prashant to discuss with Manuela before confirming his role in modelling for and developing the MIS.
 * Assuming Prashant gets okay to move forward, Wolf to facilitate integration of Prashant into the team.
 * Make available Sebastian's thesis.
 * Facilitate working with Yixin on customer preference modelling when she's back from China.
 * Further discussion on non-utility-maximing behavioral factors that feed into customer models with Alok Gupta.
 * Prashant to organize meeting between John, Wolf, Prashant and Manuela in about 4 weeks.

Follow-ups

 * 1) August 28, 2010: Discussion with Manuela complete. Okay to move forward. Suggested initial focus on (i) modelling customer behavors, (ii) analysis of which information should be available to which brokers at what time, and (iii) the representation of information stored by the MIS and communicated between the MIS and the broker agents.

[Prashant thanks Wolf and John for taking the time for this meeting.]