Searcharxiv⌕ Search

arXiv subjects

Zhijian Wang

Publications and source records attributed to Zhijian Wang.

30 records · Page 2Linked to original sources
arXiv2014

Periodic frequencies of the cycles in $2\times2$ games: evidence from experimental economics

Evolutionary dynamics provides an iconic relationship --- the periodic frequency of a game is determined by the payoff matrix of the game. This paper reports the first experimental evidence to demonstrate this relationship. Evidence comes from two populations randomly-matched $2\times2$ games with 12 different payoff matrix parameters. The directions, frequencies and changes in the radius of the cycles are measured definitively. The main finding is that the observed periodic frequencies of the persistent cycles are significantly different in games with different parameters. Two replicator dynamics, standard and adjusted, are employed as predictors for the periodic frequency. Interestingly, both of the models could infer the difference of the observed frequencies well. The experimental frequencies linearly, positively and significantly relate to the theoretical frequencies, but the adjusted model performs slightly better.

nlin.AO↗
arXiv2013

A Comment on "Cycles and Instability in a Rock-Paper-Scissors Population Game: A Continuous Time Experiment"

The authors (Cason, Friedman and Hopkins, Reviews of Economics Studies, 2014) claimed a result that the treatments (using simultaneous matching in discrete time) replicate previous results that exhibit weak or no cycles. After correct two mathematical mistakes in their cycles tripwire algorithm, we research the cycles by scanning the tripwire in the full strategy space of the games and we find significant cycles missed by the authors. So we suggest that, all of the treatments (using simultaneous matching in discrete time) exhibit significant cycles.

stat.ME↗
arXiv2013

Cyclic motions in Dekel-Scotchmer Game Experiments

TASP (Time Average Shapley Polygon, Bena{\=ı}m, Hofbauer and Hopkins, \emph{Journal of Economic Theory}, 2009), as a novel evolutionary dynamics model, predicts that a game could converge to cycles instead of fix points (Nash equilibria). To verify TASP theory, using the four strategy Dekel-Scotchmer games (Dekel and Scotchmer, \emph{Journal of Economic Theory}, 1992), four experiments were conducted (Cason, Friedman and Hopkins, \emph{Journal of Economic Theory}, 2010), in which, however, reported no evidence of cycles (Cason, Friedman and Hopkins, \emph{The Review of Economic Studies}, 2013). We reanalysis the four experiment data by testing the stochastic averaging of angular momentum in period-by-period transitions of the social state. We find, the existence of persistent cycles in Dekel-Scotchmer game can be confirmed. On the cycles, the predictions from evolutionary models had been supported by the four experiments.

stat.AP↗
arXiv2013

Cycle frequency in standard Rock-Paper-Scissors games: Evidence from experimental economics

The Rock-Paper-Scissors (RPS) game is a widely used model system in game theory. Evolutionary game theory predicts the existence of persistent cycles in the evolutionary trajectories of the RPS game, but experimental evidence has remained to be rather weak. In this work we performed laboratory experiments on the RPS game and analyzed the social-state evolutionary trajectories of twelve populations of N=6 players. We found strong evidence supporting the existence of persistent cycles. The mean cycling frequency was measured to be $0.029 \pm 0.009$ period per experimental round. Our experimental observations can be quantitatively explained by a simple non-equilibrium model, namely the discrete-time logit dynamical process with a noise parameter. Our work therefore favors the evolutionary game theory over the classical game theory for describing the dynamical behavior of the RPS game.

nlin.AO↗
arXiv2012

Test MaxEnt in Social Strategy Transitions with Experimental Two-Person Constant Sum 2$\times$2 Games

By using laboratory experimental data, we test the uncertainty of social strategy transitions in various competing environments of fixed paired two-person constant sum $2 \times 2$ games. It firstly shows that, the distributions of social strategy transitions are not erratic but obey the principle of the maximum entropy (MaxEnt). This finding indicates that human subject social systems and natural systems could have wider common backgrounds.

stat.ME↗
arXiv2012

Do cycles dissipate when subjects must choose simultaneously?

This question is raised by Cason, Friedman and Hopkins (CFH, 2012) after they firstly found and indexed quantitatively the cycles in a continuous time experiment. To answer this question, we use the data from standard RPS experiment. Our experiments are of the traditional setting - in each of repeated rounds, the subjects are paired with random matching, using pure strategy and must choose simultaneously, and after each round, each subject obtains only private information. This economics environment is a decartelized and low-information one. Using the cycle rotation indexes (CRI, developed by CFH) method, we find, the cycles not only exist but also persist in our experiment. Meanwhile, the cycles' direction are consistent with 'standard' learning models. That is the answer to the CHF question: Cycles do not dissipate in the simultaneously choose game. In addtion, we discuss three questions (1) why significant cycles are uneasy to be obtained in traditional setting experiments; (2) why CRI can be an iconic indexing-method for 'standard' evolution dynamics; and (3) where more cycles could be expected.

nlin.AO↗
arXiv2012

Evolutionary Rotation in Switching Incentive Zero-Sum Games

In a laboratory experiment, round by round, individual interactions should lead to the social evolutionary rotation in population strategy state space. Successive switching the incentive parameter should lead to successive change of the rotation ---- both of its direction and its strength. In data from a switching payoff matrix experiment of extended 2x2 games (Binmore, Swierzbinski and Proulx, 2001 [1]), we find the changing of the social evolutionary rotation can be distinguished quantitatively. The evolutionary rotation can be captured by evolutionary dynamics. With eigenvalue from the Jacobian of a constrained replicator dynamics model, an interpretation for observed rotation strength is given. In addition, equality-of-populations rank test shows that relative response coefficient of a group could persist cross the switching parameter games. The data has successively been used to support Von Neumann's minimax theory. Using the old data, with observed evolutionary rotation, this report provides a new insight into evolutionary game theory and experimental social dynamics.

stat.ME↗
arXiv2012

Maxent in Experimental 2$\times$2 Population Games

In mixed strategy 2\times2 population games, the realization of maximum entropy (Maxent) is of the theoretical expectation. We evaluate this theoretical prediction in the experimental economics game data. The data includes 12 treatments and 108 experimental sessions in which the random match human subjects pairs make simultaneous strategy moves repeated 200 rounds. Main results are (1) We confirm that experimental entropy value fit the prediction from Maxent well; and (2) In small proportion samples, distributions are deviated from Maxent expectations; interesting is that, the deviated patterns are significant more concentrated. These experimental findings could enhance the understanding of social game behavior with the natural science rule --- Maxent.

stat.AP↗
arXiv2011

Test the Principle of Maximum Entropy in Constant Sum 2x2 Game:Evidence in Experimental Economics

Entropy serves as a central observable which indicates uncertainty in many chemical, thermodynamical, biological and ecological systems, and the principle of the maximum entropy (MaxEnt) is widely supported in natural science. Recently, entropy is employed to describe the social system in which human subjects are interacted with each other, but the principle of the maximum entropy has never been reported from this field empirically. By using laboratory experimental data, we test the uncertainty of strategy type in various competing environments with two person constant sum $2 \times 2$ game. Empirical evidence shows that, in this competing game environment, the outcome of human's decision-making obeys the principle of maximum entropy.

physics.soc-ph↗
arXiv2011

Measurement and Application of Entropy Production Rate in Human Subject Social Interaction Systems

This paper illustrates the measurement and the applications of the observable, entropy production rate (EPR), in human subject social interaction systems. To this end, we show (1) how to test the minimax randomization model with experimental economics' 2$\times$2 games data and with the Wimbledon Tennis data; (2) how to identify the Edgeworth price cycle in experimental market data; and (3) the relationship within EPR and motion in data. As a result, in human subject social interaction systems, EPR can be measured practically and can be employed to test models and to search for facts efficiently.

stat.ME↗
arXiv2007

Coevolution of Mercy and Altruistic Cooperation

Besides altruistic punishment and group selection, we argue that, mercy can lead to altruistic cooperation. Modeling the micro economic behavior of the mercy, with two alleles of genes (Cooperation or Defection & Mercy or No mercy) agents in a network, we present the computational simulation results in the spatiotemporal evolution game theory frame to prove the above argument. Here, mercy (or as 'Love thy neighbors') means, the agents, with mercy preference, might share his own fitness with his poorest neighbor who poorer than himself.

nlin.AO↗