Huberman, B. A., & Glance, N. S. (1993). Evolutionary games and computer simulations. Proceedings of the National Academy of Sciences, USA, 90(16), 7715--7718.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....whom to exploit a better payoff. The results of our simulations show that the network of agents reaches a steady state with a high degree of cooperative behavior. The fraction of agents that cooperate depends only slightly on the incentive to defect in the individued game. This See for e.g. [14,11,10,6,15]. A cellular automata representation with severed learn ing rules is presented in Ref. 8] For an extensive study of the dynamics of a PD game with different strategies, evolution of the strategies and different networks see Ref. 4] behavior contrasts with previous studies on spatial PD ....

....been previously studied in Refs. 14,15,13] These studies showed how partial cooperation can be sustained due to the local interactions, in spite of memoryless strategies. Several extensions to this spatial model have been studied in the literature. For example, introducing asynchronous updates [6] or introducing errors in the imitation process [12] the basic results persist [13] For comparison with the adaptive network considered below, we give in Table 2 some numerical results for a fixed (p = 0) random network. Such network is our initial condition for the adaptive network. We see ....

B. A. Huberman and N. S. Glance. Evolutionary games and computer simulations. Proc. Natl. Acad. Sci. USA, 90:7716-7718, 1993. 14 M.G. Zimmermann et ed.

....population dynamics ( Hamilton et al., 1990] Asynchronous play and reproduction. In the experiments reported here, all actions were in effect done synchronously. It is known that in some situations going from synchronous to asynchronous updates can have dramatic effects on population dynamics [Huberman and Glance, 1993]. Studies into the effects of varying the localization of information and the synchrony of updating along the lines started by [Hoffmann and Waring, 1996] should be carried out for tag based systems as well. These are just of the few possibilities for future research. In addition, careful studies ....

Huberman, Bernardo A. and Glance, Natalie S. "Evolutionary Games and Computer Simulations." Proc. Natl. Acad. Sciences (USA), 90, 7716-18.

....population dynamics ( Hamilton et al., 1990] Asynchronous play and reproduction. In the experiments reported here, all actions were in effect done synchronously. It is known that in some situations going from synchronous to asynchronous updates can have dramatic effects on population dynamics [Huberman and Glance, 1993]. Studies into the effects of varying the localization of information and the synchrony of updating along the lines started by [Hoffmann and Waring, 1996] should be carried out for tag based systems as well. These are just of the few possibilities for future research. In addition, careful studies ....

Huberman, Bernardo A. and Glance, Natalie S. "Evolutionary Games and Computer Simulations." Proc. Natl. Acad. Sci- ences (USA), 90, 7716-18.

....are not contentneutral. Their mechanisms can introduce artifacts determined more by the modeling technology than by the system being modeled. Earlier researchers have pointed out such effects within agent based models, based on differences between synchronous and asynchronous execution [2, 4]. Our results in this paper reinforce our earlier observations [8] about the loss of ontological distribution in an equation based model. Our conclusion is cautionary, not fatalistic. We do not reject system modeling and simulation as impossible. In fact, it is unavoidable in engineering control ....

B. A. Huberman and N. S. Glance. Evolutionary Games and Computer Simulations. Proceedings of the National Academy of Science USA, 90(August):7716-7718, 1993.

....of vision of 120 degrees and their maximum perception distance (MaxView) is 50 units. Parallel simulations cannot be run on most computers and therefore, activities of agents are regulated by a timing regime. Studies have shown that a specific timing regime influences the results of the simulation [33]. Often a random regime is applied in which each entity receives a random waiting time from a uniform distribution and the one with the shortest waiting time is activated first. Here, I combine a random regime with a biologically plausible timing regime that is locally controlled by other entities ....

Huberman, B.A. and N.S. Glance, Evolutionary games and computer simulations. Proceedings of the National Academy of Sciences of the United States of America, 1993. 90: p. 7716-7718.

....agent is eventually replaced by an ALLC or adaptive TFT neighbor that has achieved a higher score in its own different spatial tournament. Figure 8 gives the results of spatial irregularity in a random and a regular initial configuration. Finally, we have experimented with continuous time games ([7][9] where agents do not play against their opponents simultaneously and are not updated synchronously, but they play and are updated one after the other in random order. In this case the general conclusions of previous game models apply as well. A detailed analysis in the same spirit as that of ....

Huberman, B.A., and Glance, N.S.: Evolutionary games and computer simulations, Proceedings of the National Academy of Sciences, 90 (1993) 7712-7715

....in turn below. 10 In general, we can expect these alternative execution regimes to yield different individual agent histories, and possibly different macrosocial outcomes as will be illustrated below. Indeed, even 9 An important exception to this statement are multi agent models of traffic. In [10] a critique of the use of synchronous updating in certain game theoretic models is rendered, demonstrating that results from such models are typically not robust to asynchronous updating. 10 The case in which agents have differential incentive to be active is treated in [12] for a given type ....

Huberman, B.A. and N.S. Glance. 1993. "Evolutionary Games and Computer Simulations." Proceedings of the National Academy of Sciences USA, 90: 7716-18.

.... optimal or equilibrium solution is computed in a centralized, off line manner. Examples of this approach include [20, 112] 8 ffl Decentralized heuristic adjustment. The price of each resource is adjusted in some sensible way that simulates a market process. Examples of this approach include [54, 51, 50], 58, 67, 83] The offline calculation is attractive when there are strong interaction effects between the resources and when the allocation problem is relatively static or changes slowly. The heuristic approach is better suited to problems where there a number of relatively independent ....

B. Huberman and N. Glance. Evolutionary games and computer simulations. NAS Proceedings, 1993.

....modeling technologies are not content neutral. They can introduce artifacts determined more by the modeling technology than by the system being modeled. Earlier researchers have pointed out such effects within agent based models, based on differences between synchronous and asynchronous execution [9, 11]. Our results in this paper reinforce our earlier observations [16] about the loss of ontological distribution in an equation based model. System modeling and simulation are not impossible. In fact, they are unavoidable in engineering agent based systems, due to the analytical intractability of ....

B. A. Huberman and N. S. Glance. Evolutionary Games and Computer Simulations. Proceedings of the National Academy of Science USA, 90(August):7716-7718, 1993.

....a number of iterations. Similar phenomena occur in backpropagation neural networks [18] These networks are often used to model the learning behaviour of agents. 8 5. Modelling Time A wide spread report on alignment by Huberman and Glance shows the importance of modelling the passing of time [19]. The authors replicate a simulation purporting to produce spatial patterns of co operate and defect moves in a Prisoner s Dilemma, using synchronous updates. Since Huberman and Glance reason that synchronous updates are rare in the real world, they move to asynchronous updates. Their simulation ....

Bernardo A. Huberman and Natalie S. Glance. Evolutionary games and computer simulations. Proc National Academy of Sciences USA, pages 7716--7718, 1993.

....Similarly, this work considers multi agent learning in positive sum games. The results described here on convergence to Stackelberg equilibria, rather than traditional game theoretic solutions such as D 1 , were largely a result of the asynchronous nature of play. Likewise, Huberman and Glance [28] simulate the evolutionary dynamics of both synchronous and asynchronous updating in the repeated Prisoners Dilemma, and nd that rather di erent behaviors arise. Perhaps of closer interest to the present study, empirical investigations of asynchronous Internet interactions were reported by ....

B.A. Huberman and N.S. Glance. Evolutionary games and computer simulations. Proceedings of the National Academy of Sciences USA, 90:7716-7718, 1993.

....interesting implications regarding the polymorphic conviviality of cooperators and defectors, depend critically on the synchronous updating scheme they use. When asynchrony is introduced no spatial pattern appears, and the much gloomier picture of global defection as the stable strategy results, (Huberman Glance, 1993) 1 . But this is not the only example from which a similar lesson can be learned. Abramson and Zanette (1998) show that asyn 1 In (May, Bonhoe er, Nowak, 1995) the original choice of synchronous updating is defended by saying that it may be appropriate for some biological situations. This ....

Huberman, B. A., & Glance, N. S. (1993). Evolutionary games and computer simulations. Proc. Natl. Acad. Sci. USA, 90, 7715-7718.

....among auctions is a feature that is buried in the continuous version of the tatonnement process. Although it is usually not stressed, in the translation from a difference equation formalization of tatonnement to a differential formalization, the simultaneity (among auctions) is lost (see Huberman and Glance (1993) for a discussion) 4 The current implementation of walras chooses with equal probability from the set f0; 1; 2g. Note that with this setting, each agent submits an average of one bid per cycle. As we see below, the particular distribution is inessential to our analysis. walconv.tex; 12 07 1995; ....

Huberman, B. A., and Glance, N. S. 1993. Evolutionary games and computer simulations. In Proceedings of the National Academy of Sciences.

.... emergent phenomena and thus enhance observability, while subsequent phases of the study will certainly involve continuous time scales (or, equivalently, delayed action models) which have been shown to have drastic impact on the dynamic behavior and especially the stability of distributed systems (Huberman Glance (1993), Bersini Detours (1994) The results presented next in increasing order of example complexity do not purport to be significant or novel by themselves, but to demonstrate through seemingly simple examples the potential of the model to account for complex emergent phenomena. This is also the ....

Huberman, B.A., N.S. Glance (1993). Evolutionary games and computer simulations, Proceedings National Academy of Sciences (USA), Vol. 90, pp. 7716-7718, 1993.

....yield several significant advances. To date, research in multiple agent planning and control has been limited largely to the area of distributed artificial intelligence (Rosenschein and Genesereth 1985; Stone and Veloso 1996a; Suguwara and Lesser 1993; Tan 1993) and artificial life (Collins 1992; Huberman and Glance 1995; Sandholm and Crites 1995; Stanley, Ashlock, and Tesfastsion 1993) In distributed AI (DAI) several agents cooperate to achieve some goal or accomplish some task. The task is usually one of sufficient complexity that no single agent can accomplish the task alone. Because the agents cooperate, ....

Huberman, B. and Glance, N. (1995). Evolutionary games and computer simulations.

....but agents will switch only if they perceive a gain in personal utility. This is only one of many ways to simulate such a model. Our experience running similar types of simulations indicates that the most important feature is that the agent and managerial states be updated asynchronously [10], not synchronously, in order to accurately model continuous time. Other features of the algorithm described above have little effect on the qualitative results described in the next section. Results The dynamics on the organizational level mirrors the agent level description given earlier: when ....

Bernardo A. Huberman and Natalie S. Glance. Evolutionary games and computer simulations. Proceedings of the National Academy of Science USA, 90:7716-- 7718, August 1993.

No context found.

Huberman, B. A., & Glance, N. S. (1993). Evolutionary games and computer simulations. Proceedings of the National Academy of Sciences, USA, 90(16), 7715--7718.

No context found.

Huberman, B.A., Glance, N.S., 1993. Evolutionary games and computer simulations. Proc. Natl.

No context found.

HUBERMAN, B. A., AND N. S. GLANCE (1993): "Evolutionary Games and Computer Simulations," Proc.

No context found.

B. A. Huberman and N. S. Glance. Evolutionary games and computer simulations. Proc. Natl. Acad. Sci. USA, 90:7716--8, 1993.

No context found.

B. Huberman and N. Glance. Evolutionary games and computer simulations. Proceedings of the National Academy of Science, 90:7716-7718, 1993.

No context found.

Huberman, B.A., and N. S. Glance (1993), "Evolutionary Games and Computer Simulations," Proceedings of the National Academy of Sciences, 90 (6), pp. 7716-7718.

No context found.

B. A. Huberman and N. S. Glance. Evolutionary Games and Computer Simulations. Proceedings of the National Academy of Science USA, 90(August):7716-7718, 1993.

No context found.

Huberman, B.A., Glance, N.S.: Evolutionary games and computer simulations. In: Proceedings of the National Academy of Science USA 90. (1993) 7716--18

No context found.

B. A. Huberman, N.S. Glance, "Evolutionary games and computer simulations," Proceedings of National Academy of Sciences of the USA, 90, pp. 7716-7718, 1993.

First 50 documents

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC