Poster Session and Reception

Monday, November 3, 2003 - 4:30pm - 6:00pm
Lind 400
  • Urban Infrastructure Suite: An integrated Simulation Based Analytical Tool for Critical Infrastructures
    Stephen Eubank (Los Alamos National Laboratory)
  • Measurement and Analysis of Large Social and Infrastructure Networks
    Madhav Marathe (Los Alamos National Laboratory)
  • Mathematical and Computational Theory of Sequential Dynamical Systems
    Henning Mortveit (Los Alamos National Laboratory)
  • CSIRO's Agent-Based Modelling Working Group
    David Batten (Commonwealth Scientific and Industrial Research Organization (CSIRO))
    Australia's largest research organization, CSIRO (Commonwealth Scientific and Industrial Research Organization) established a Centre for Complex Systems Science (CSS) two years ago. Within this CSS Centre, there is a small suite of agent-based modelling projects embracing a broad range of contexts -- from electricity markets to fishing behaviour, rangelands and river catchments. The CSIRO Agent-based Modelling Working Group nurtures each of these projects by facilitating interaction tasks such as regular workshops and working group meetings. International experts from Europe and North America also attend. The Coordinator of this Working Group, Dr. David Batten ( is directing the ABM work on Australia's electricity market, but he would be pleased to discuss any of the Working Group's ABM projects with interested parties. Collaboration with research groups engaged in similar research in other countries would be especially welcome.
  • Scheduling Tasks with Precedence Constraints to Solicit Desirable Bid Combinations
    Maria Gini (University of Minnesota, Twin Cities)
    Joint work with Wolfgang Ketter and John Collins.

    We study the problem of optimizing the time windows in Requests for Quotes that an agent sends to other agents to obtain bids for combinations of tasks with complex time constraints and interdependencies. Our approach uses Expected Utility Theory to reduce the likelihood of receiving unattractive bids, while maximizing the number of bids that are likely to be included in the winning bundle. We describe the model, illustrate its operation and properties, and discuss what assumptions are required for its successful integration into multi-agent applications.
  • Simulating Nanorobots in Viscous Fluids
    Tad Hogg (HP (Hewlett-Packard))
    Joint work with Adriano Cavalcanti.

    Developing nanoscale robots (nanorobots) presents difficult fabrication and control challenges [4]. Of particular interest are medical applications [2] in which the robots operate in fluid microenvironments in the body. While such robots cannot yet be fabricated, theoretical and simulation studies identify plausible designs and capabilities [1,2].

    To aid investigation of system-level control algorithms for these robots, we present a physically-based simulator for nanorobots in a simplified fluid environment motivated by medically relevant microenvironments. The robots' motions, characterized by a low Reynolds number, are quite different from common experience with larger, faster flows [3].

    [1] K. E. Drexler, Nanosystems, Wiley 1992

    [2] R. A. Freitas Jr., Nanomedicine, vol. 1, Landes Bioscience, 1999, at

    [3] E. M. Purcell, Life at Low Reynolds Number, American Journal of Physics, 45:3-11 (1977)

    [4] A. A. G. Requicha, Nanorobots, NEMS and Nanoassembly, to appear in Proc. of IEEE special issue on Nanoelectronics and Nanoprocessing
  • An Evolutionary Framework for Studying Behaviors of Economic Agents
    Wolfgang Ketter (University of Minnesota, Twin Cities)
    Agents that interact in electronic marketplaces. We describe how this approach can be used when agents' strategies are based on different methodologies, employing incompatible rules for collecting information and for reproduction. We present experimental results from a simulated market, where multiple service providers compete for customers using different deployment and pricing schemes. The results show that heterogeneous strategies evolve in the same market and provide useful research data.
  • The Emergence of Temporal Correlations in a Study of Global Economic Interdependence
    Adam Landsberg (Claremont McKenna College)
    We develop a simple firm-based automaton model for global economic interdependence of countries using modern notions of self-organized criticality and dynamical renormalization group methods. We demonstrate how extremely strong statistical correlations can naturally develop between two countries even if the financial interconnections between those countries remain very weak. Potential policy implications of this result are also discussed.

    Joint work with Eric J. Friedman (Cornell University) and Simon Johnson (MIT)
  • A Computational Algebra Algorithm for Reverse-engineering of Gene Regulatory Networks
    Reinhard Laubenbacher (Virginia Polytechnic Institute and State University)
    One of the central problems in systems biology is to model gene regulatory networks from experimental data. Several modeling frameworks have been proposed, that can be categorized broadly into static versus dynamic, continuous versus discrete, and deterministic versus stochastic. We present a method that infers a multi-state, discrete dynamic network from one or more time series of DNA microarray measurements. The method utilizes algorithms from computational algebra and algebraic geometry. We validate our reverse-engineering algorithm using simulated data generated by a Boolean network model of the regulatory network responsible for pattern formation in Drosophila melanogaster.
  • Cooperation in an Unpredictable Environment
    Kristian Lindgren (Chalmers University of Technology/University of Göteborg)
    Joint work with Anders Eriksson.

    One of the main limitations with the Prisoner's Dilemma game and many of the similar games studied is the static character of the interaction situation, i.e that players always have the same payoff elements in all encounters. In practice it is much more common that circumstances change over time, so that the payoff elements are rarely the same at any two encounters.

    We investigate the evolution of cooperation in a random environment, when players engage in repeated interactions. Strategies are represented as a small set of behavioral states, with transitions between them.

    For very low mistakes rates, the level of cooperation is high but unstable. A qualitative model is introduced to show that the fluctuations are due the breakdown of reciprocation at high levels of cooperation. At higher mistakes rates, the genetic drift is decreased significantly.

    We conclude that also with players with very limited faculties, it is possible to establish robust cooperation in the presence of uncertain future payoffs. The possibility of making mistakes makes it harder to establish robust cooperation, but on the other hand it is not as susceptible to genetic drift.
  • A Scale-Free Network Model of Urban Economics
    Kristian Lindgren (Chalmers University of Technology/University of Göteborg)
    Joint work with Claes Andersson and Alexander Hellervik.

    The geographic distribution of human land use exhibits a wide range of large-scale regularities with maybe the most widely known example being the rank-size rule, also known as Zipf's Law. This rule states that a city's size and rank (1=largest, 2= second largest etc) has a power law relation. Many properties of the urban system, in addition to the rank-size rule, exhibit scaling over a considerable number of orders of magnitude, i.e. land value per land area, land value per city and the relationship between urban area and perimeter. Although the fractal nature of the urban system has been known since long, explanations as to what processes are responsible for this behavior are at best incomplete. We have developed a complex network model where nodes are taken to be fixed-size non-overlapping lots of land and connections are trade relations between economic activities in these lots. Through a connection to theory on land markets we can compare node degrees with empirically observed land value data. We obtain excellent agreement on several levels: land value per unit area, land value per city and the relationship between city area and perimeter. In addition we also show emprically that there is a linear relation between land value and population, thus making our results directly applicable to Zipf's Law of city sizes.
  • Agent Based Modeling Approach to Self-Organizing Neural Networks
    Timothy Schoenharl (University of Notre Dame)
    We explore a novel self-organizing neural network topology. We have drawn inspiration from recent research into complex networks and advances in neurobiology, and applied it towards the development of an artificial neural network. An agent based approach is used to model the neurons and their connections, providing a richness of expression not available in other neural network simulations. The agent based paradigm is well suited for our exploration, as local interaction among the neurons drives the evolution of the global network topology. We demonstrate our simulation and discuss results.
  • Using Cultural Algorithms to Evolve Strategies for Recessionary Markets
    David Ostrowski (Ford Motor Company)

    Cultural Algorithms are computational self-adaptive models which utilize a population and a belief space. In this framework, a white and black box testing strategy is embedded in order to test large-scale GP programs. The model consists of two populations, one supporting white box testing of a genetic programming system and the other supporting black box testing. The two populations communicate with each other by means of a shared belief space. This is applied to the calibration of a multi-agent system by allowing for evolution of near optimal parameters. The Cultural approach is employed to abstract coefficients of pricing strategies that are applied to a complex model of durable goods. This model simulates consumer behaviors as applied in the context of economic cycles.


    Agent based techniques are known to complement standard economic theory [Axtell] Due to mathematical tractability constraints, an evolutionary framework can be successful in terms of being able to derive a solution. We have employed the utilization of a multi-agent system, MarketScape, to simulate a real-world consumer market. When we apply heterogeneous factors to the application of this market, it can be demonstrated that traditional economic theory does not hold. [Tassier] An example of this is where the postpone scenario in which consumers will delay purchase of economic goods as a function of past prices and time [Tassier] . This has been noted as affecting purchase behavior. Specific strategies by that of OEMs such as the placement of incentives are demonstrated to actually bring about a decrease in profitability. Another example involves the application of memory based techniques to that of market recession.

    Here, we are interested in the application of Software Engineering techniques to that of the calibration of agent-based design. Complementary techniques of White and Black box testing are demonstrated to assist in the efficient design of software. [Ostrowski] In order to accomplish this, the White Box approach is applied to consider the implementation of specific requirements in order to guide the initial and subsequent design process. The assumption in this approach is that the results of testing can be used to guide the search for a program.


    [1] R. Axtell. Why agents? On the varied motivations for agent computing in the social sciences, 2001. mimeo, Center on Social and Economic Dynamics , The Brookings Institute.

    [2] R. Porter, P.Sattler. Patterns of trade in the market for used durables: Theory and evidence, 1999 NBER Working Paper No W7419.

    [3] Sommerviille, I., Software Engineering, Addison-Wesley, 1996.

    [4] R.G Reynolds, An Introduction to Cultural Algorithms, In the Proceedings of the 3rd annual Conference on Evolution Programming , Sebalk, A.V. Fogel L.J., River Edge, NJ. World Sientific Publishing, 1994, pp 131-13.

    [5] Chung, Chan-Jin, Reynolds. R.G, A Testbed for Solving Optimization Problems Using Cultural Algorithmns, In proceedings of the fifth Annual Conference on Evolutionary Programing, MIT Press, 1996.

    [6] Chung, Chan-Jin, Reynolds, R.G., Knowledge-Based Self-Adaptation Using Cultural Algorithms: An Evolutionary Programming Example, International Journal on Artificial Intelligence Tools, Vol. 7, No.3, September 1998,pp. 239-293.

    [7] Rychtyckyj N Reynolds, Using Cultural Algorithms to Improve Performance in Semantic Networks, Angeline Peter, CEC99.

    [8] E. Zannoni, Cultural Algorithms with Genetic Programmig: Learning to Control the Program Evolution Process. Phd. Thesis, Computer Science, Wayne State University , Detroit 1996.

    [9] D. Ostrowski, R.G. Reynold