2012 Courses

Prof. Barry G. Silverman, 120 Hayden Hall, (215) 573-8368
Office Hours: Tuesday and Thursday: 2 – 4 PM

basil@seas.upenn.edu

ESE 308 Abstract

AGENT-Based Modeling & Simulation:

Artificial Life, Human Behavior, and Socio-Technical Systems

Classroom: Towne 305, Labs in Moore 100B

Agents are a new technique for trying to model, simulate, and understand systems that are ill-structured and whose mathematics is initially unknown and possibly unknowable. This approach allows the analyst to assemble models of agents and components where micro-decision rules may be understood; to bring the agents and components together as a system where macro-behavior then emerges; and to use that to empirically probe and improve understanding of the whole, the interrelations of the components, and synergies. This approach helps one explore parametrics, causality, and what-ifs about socio-technical systems (technologies that must support people, groups, crowds, organizations, and societies). It is applicable when trying to model and understand human behavior – consumers, investors, passengers, plant operators, patients, voters, political leaders, terrorists, and so on. This course will allow students to investigate and compare increasingly complex agent based paradigms along three lines – math foundations, heuristic algorithms/knowledge representations, and empirical science. The student will gain a toolbox and methodology for attempting to represent and study complex socio-technical systems. Prereqs: probability, Java or C programming, or equivalent.

Goals:

Students taking this class will learn to:

• Understand the different paradigms of the agent based approach (eg., large numbers of “light agents”, finite state machines, game theoretic rational actor theory, and use of richly detailed socio-cognitive and affective agents) and when they each can be helpful.
• Flowchart, design, build, and evaluate your own agent model of a system with particular emphasis on decision science methods for agent choice under certainty, uncertainty, and risk.
• Use the agent approach to model artificial life, artificial societies, and complex systems problems.
• Conduct modeling and simulation studies to analyze the robustness and emergence of system equilibria as a function of sensitivity to parameter shifts and policy changes.

Textbook: 

Michael D. Resnik (1987)  Choices: An Introduction to Decision Theory.  Minneapolis: University of Minnesota Press.

Packet of Readings about Agent Based Modeling and Simulation

COURSE ABSTRACT

ESE 590 –Systems Methodology

Winter

Classroom: Moore 216 and Labs in Moore 100B

Professor: Dr. Barry G. Silverman, 120 Hayden Hall, (215) 573-8368

Catalog Description: This course covers the methodologies and techniques important to DESIGNING large complex, purposeful systems and to discovering policies that influence them throughout the stages of their lifecycle. The course focuses on hands-on synthetic thinking, where students assemble the big picture from modeling the individual actors, organizations, and artifacts in a socio-technical system of interest. This is the study of emergence of macro-behavior from the micro-decision making of the actors involved - to inquire into the design of a purposeful system, and to examine alternative futures that are ideal, yet affordable, sustainable, and workable. Specifically, the student learns systems theory, systems methodologies (design inquiry/learning systems, idealized design/interactive planning, and soft systems methodology/knowledge management), bottom up modeling (decision science, multi-attribute utility theory, affective reasoning, agent based modeling, simulated societies), and how to further research and apply the synthetic paradigm.

 Course Objectives: What you should know and skills you should have by the end of this course:

• To understand the differences between reductive/analytical science and the synthetic SE paradigm. To shift from focusing on the parts to focusing on wholes, on relations between teleological sub-systems, and on synergy and emergence.
• To accrue indepth knowledge and experience (via repeated case analysis) of at least one standard SE methodology and how to use it to formulate and solve complex socio-technical system problems. [Part I ]
• To learn how to apply decision theories and toolsets to help elicit system goals/requirements and (leading) performance indices by which a system’s improvement can be measured and service disruptions can be minimized. [Part I, II, IV]
• To practice hands-on with model-based systems engineering in general (and agent based modeling and simulation tools in particular): (a) to study a complex system’s life cycle, feedback loops, costs, “ilities” (reliability, maintainability, sustainability), and equilibria tipping/emergence, and (b) to explore ways to improve these items. [Part II]
• To gain a repertoire of different SE Methodologies, plus insight into each methodology’s pros and cons and when it should be used (or not). [Part III]
• To develop know-how and a continual learning attitude about potential system disruptions and about proactive/emergent resilience enhancement practices to help avoid, survive, and recover from accidents and failure modes (Part IV).