Parrondo games with spatial dependence, II

arXiv1206.6195MaRDI QIDQ6233977FDOQ6233977

Publication date: 27 June 2012

Abstract: Let game B be Toral's cooperative Parrondo game with (one-dimensional) spatial dependence, parameterized by N (3 or more) and p_0, p_1, p_2, p_3 in [0,1], and let game A be the special case p_0=p_1=p_2=p_3=1/2. In previous work we investigated mu_B and mu_(1/2,1/2), the mean profits per turn to the ensemble of N players always playing game B and always playing the randomly mixed game (1/2)(A+B). These means were computable for N=3,4,5,...,19, at least, and appeared to converge as N approaches infinity, suggesting that the Parrondo region (i.e., the region in which mu_B is nonpositive and mu_(1/2,1/2) is positive) has nonzero volume in the limit. The convergence was established under certain conditions, and the limits were expressed in terms of a parameterized spin system on the one-dimensional integer lattice. In this paper we replace the random mixture with the nonrandom periodic pattern A^r B^s, where r and s are positive integers. We show that mu_[r,s], the mean profit per turn to the ensemble of N players repeatedly playing the pattern A^r B^s, is computable for N=3,4,5,...,18 and r+s=2,3,4, at least, and appears to converge as N approaches infinity, albeit more slowly than in the random-mixture case. Again this suggests that the Parrondo region (mu_B is nonpositive and mu_[r,s] is positive) has nonzero volume in the limit. Moreover, we can prove this convergence under certain conditions and identify the limits.

Mathematics Subject Classification ID

Applications of Markov chains and discrete-time Markov processes on general state spaces (social mobility, learning theory, industrial processes, etc.) (60J20) Interacting random processes; statistical mechanics type models; percolation theory (60K35)

This page was built for publication: Parrondo games with spatial dependence, II

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6233977)