Dynamics of computational ecosystems

Kephart, J. O.; Hogg, Tad; Huberman, Bernardo A.

doi:10.1103/physreva.40.404

Cited by 123 publications

(61 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…(10) for the case in which agents choose among two resources with cooperative payoffs, a case which we have shown to generate chaotic behavior in the absence of rewards [5,6]. As in the particular example of Fig.…”

Section: Resultsmentioning

confidence: 99%

Dynamics of Large Autonomous Computational Systems

Hogg

Huberman

2004

Collectives and the Design of Complex Systems

View full text Add to dashboard Cite

distributed systemsDistributed large scale computation gives rise to a wide range of behaviors, from the simple to the chaotic. This diversity of behaviors stems from the fact that the agents and programs have incomplete knowledge and imperfect information on the state of the system. We describe an instantiation of such systems based on market mechanisms which provides an interesting example of autonomous control. We also show that when agents choose among several resources, the dynamics of the system can be oscillatory and even chaotic. Furthermore, we describe a mechanism for achieving global stability through local controls.

show abstract

Section: Resultsmentioning

confidence: 99%

Dynamics of Large Autonomous Computational Systems

Hogg

Huberman

2004

Collectives and the Design of Complex Systems

View full text Add to dashboard Cite

show abstract

“…14. Kephart et al (Kephart, Hogg, & Huberman, 1989) describe another setting where sophisticated agents that try to anticipate the actions of others often make results worse for themselves. In this model, the sophisticated agents' downfall is their failure to account properly for simultaneous adaptation by the other agents.…”

Section: Theorem 6 Let U Be a Continuous Strictly Concave Function Omentioning

confidence: 99%

Untitled

Wellman

1998

Machine Learning

View full text Add to dashboard Cite

Abstract. Learning in a multiagent environment is complicated by the fact that as other agents learn, the environment effectively changes. Moreover, other agents' actions are often not directly observable, and the actions taken by the learning agent can strongly bias which range of behaviors are encountered. We define the concept of a conjectural equilibrium, where all agents' expectations are realized, and each agent responds optimally to its expectations. We present a generic multiagent exchange situation, in which competitive behavior constitutes a conjectural equilibrium. We then introduce an agent that executes a more sophisticated strategic learning strategy, building a model of the response of other agents. We find that the system reliably converges to a conjectural equilibrium, but that the final result achieved is highly sensitive to initial belief. In essence, the strategic learner's actions tend to fulfill its expectations. Depending on the starting point, the agent may be better or worse off than had it not attempted to learn a model of the other agents at all.

show abstract

“…To study the effectiveness of this mechanism for controlling chaos, we examined the dynamics for the case which we had earlier shown to generate chaotic behavior in the absence of rewards (Huberman and Hogg, 1988;Kephart, Hogg, and Huberman, 1989). In this case, agents choose among two resources with cooperative payoffs; oscillations can occur as cooperation rises and fragments.…”

Section: Equilibrium Through the Use Of Profitsmentioning

confidence: 99%

Distributed Computation as an Economic System

Huberman

Hogg

1995

Journal of Economic Perspectives

View full text Add to dashboard Cite

A s everyone knows, computer networks play an increasingly important role in the economy: in the daily operation of big firms, the smooth running of government agencies, and transactions in financial markets. However, it may surprise even professional economists to find that economics is of help in allocating computational resources to programs that run in these systems.The reason is that computer networks can be regarded as a community of concurrent processes, or a "computational ecosystem" (Huberman and Hogg, 1988). As these networks grow, they become analogous to human market economies because the processes-in their interactions, strategies, and lack of perfect knowledge-face the same issues as people in a market. This paper will explain this analogy and explore one of its main implications: that economics may offer new ways to design and understand the behavior of these emerging computational systems. In particular, markets provide examples of methods to deal successfully with coordinating asynchronous operations in the face of imperfect knowledge. A computational system set up along market rules can allow the system as a whole to adapt to changes in the environment or disturbances to individual members. In what follows we will describe how market-based computational systems are already working and how they may sometimes break down. Some of the best-known programs already using these ideas include computational resource allocation (Malone et al., 1988;Waldspurger et al., 1992;Nash, 1993) and thermal markets for controlling building environments (Clearwater and Huberman, 1993), which have shown increased performance when compared with traditional operating systems.

show abstract

Dynamics of computational ecosystems

Cited by 123 publications

References 9 publications

Dynamics of Large Autonomous Computational Systems

Dynamics of Large Autonomous Computational Systems

Untitled

Distributed Computation as an Economic System

Contact Info

Product

Resources

About