We study a symmetric generalization $mathfrak{p}^{(N)}_k(eta, alpha)$ of the binomial distribution recently introduced by Bergeron et al, where $eta in [0,1]$ denotes the win probability, and $alpha$ is a positive parameter. This generalization is ba
sed on $q$-exponential generating functions ($e_{q^{gen}}^z equiv [1+(1-q^{gen})z]^{1/(1-q^{gen})};,e_{1}^z=e^z)$ where $q^{gen}=1+1/alpha$. The numerical calculation of the probability distribution function of the number of wins $k$, related to the number of realizations $N$, strongly approaches a discrete $q^{disc}$-Gaussian distribution, for win-loss equiprobability (i.e., $eta=1/2$) and all values of $alpha$. Asymptotic $Nto infty$ distribution is in fact a $q^{att}$-Gaussian $e_{q^{att}}^{-beta z^2}$, where $q^{att}=1-2/(alpha-2)$ and $beta=(2alpha-4)$. The behavior of the scaled quantity $k/N^gamma$ is discussed as well. For $gamma<1$, a large-deviation-like property showing a $q^{ldl}$-exponential decay is found, where $q^{ldl}=1+1/(etaalpha)$. For $eta=1/2$, $q^{ldl}$ and $q^{att}$ are related through $1/(q^{ldl}-1)+1/(q^{att}-1)=1$, $forall alpha$. For $gamma=1$, the law of large numbers is violated, and we consistently study the large-deviations with respect to the probability of the $Ntoinfty$ limit distribution, yielding a power law, although not exactly a $q^{LD}$-exponential decay. All $q$-statistical parameters which emerge are univocally defined by $(eta, alpha)$. Finally we discuss the analytical connection with the P{o}lya urn problem.
The paper that is commented by Touchette contains a computational study which opens the door to a desirable generalization of the standard large deviation theory (applicable to a set of $N$ nearly independent random variables) to systems belonging to
a special, though ubiquitous, class of strong correlations. It focuses on three inter-related aspects, namely (i) we exhibit strong numerical indications which suggest that the standard exponential probability law is asymptotically replaced by a power-law as its dominant term for large $N$; (ii) the subdominant term appears to be consistent with the $q$-exponential behavior typical of systems following $q$-statistics, thus reinforcing the thermodynamically extensive entropic nature of the exponent of the $q$-exponential, basically $N$ times the $q$-generalized rate function; (iii) the class of strong correlations that we have focused on corresponds to attractors in the sense of the Central Limit Theorem which are $Q$-Gaussian distributions (in principle $1 < Q < 3$), which relevantly differ from (symmetric) Levy distributions, with the unique exception of Cauchy-Lorentz distributions (which correspond to $Q = 2$), where they coincide, as well known. In his Comment, Touchette has agreeably discussed point (i), but, unfortunately, points (ii) and (iii) have, as we detail here, visibly escaped to his analysis. Consequently, his conclusion claiming the absence of special connection with $q$-exponentials is unjustified.
The theory of large deviations constitutes a mathematical cornerstone in the foundations of Boltzmann-Gibbs statistical mechanics, based on the additive entropy $S_{BG}=- k_Bsum_{i=1}^W p_i ln p_i$. Its optimization under appropriate constraints yiel
ds the celebrated BG weight $e^{-beta E_i}$. An elementary large-deviation connection is provided by $N$ independent binary variables, which, in the $Ntoinfty$ limit yields a Gaussian distribution. The probability of having $n e N/2$ out of $N$ throws is governed by the exponential decay $e^{-N r}$, where the rate function $r$ is directly related to the relative BG entropy. To deal with a wide class of complex systems, nonextensive statistical mechanics has been proposed, based on the nonadditive entropy $S_q=k_Bfrac{1- sum_{i=1}^W p_i^q}{q-1}$ ($q in {cal R}; ,S_1=S_{BG}$). Its optimization yields the generalized weight $e_q^{-beta_q E_i}$ ($e_q^z equiv [1+(1-q)z]^{1/(1-q)};,e_1^z=e^z)$. We numerically study large deviations for a strongly correlated model which depends on the indices $Q in [1,2)$ and $gamma in (0,1)$. This model provides, in the $Ntoinfty$ limit ($forall gamma$), $Q$-Gaussian distributions, ubiquitously observed in nature ($Qto 1$ recovers the independent binary model). We show that its corresponding large deviations are governed by $e_q^{-N r_q}$ ($propto 1/N^{1/(q-1)}$ if $q>1$) where $q= frac{Q-1}{gamma (3-Q)}+1 ge 1$. This $q$-generalized illustration opens wide the door towards a desirable large-deviation foundation of nonextensive statistical mechanics.
We study chaotic orbits of conservative low--dimensional maps and present numerical results showing that the probability density functions (pdfs) of the sum of $N$ iterates in the large $N$ limit exhibit very interesting time-evolving statistics. In
some cases where the chaotic layers are thin and the (positive) maximal Lyapunov exponent is small, long--lasting quasi--stationary states (QSS) are found, whose pdfs appear to converge to $q$--Gaussians associated with nonextensive statistical mechanics. More generally, however, as $N$ increases, the pdfs describe a sequence of QSS that pass from a $q$--Gaussian to an exponential shape and ultimately tend to a true Gaussian, as orbits diffuse to larger chaotic domains and the phase space dynamics becomes more uniformly ergodic.
We introduce a new universality class of one-dimensional unimodal dissipative maps. The new family, from now on referred to as the ($z_1,z_2$)-{it logarithmic map}, corresponds to a generalization of the $z$-logistic map. The Feigenbaum-like constant
s of these maps are determined. It has been recently shown that the probability density of sums of iterates at the edge of chaos of the $z$-logistic map is numerically consistent with a $q$-Gaussian, the distribution which, under appropriate constraints, optimizes the nonadditive entropy $S_q$. We focus here on the presently generalized maps to check whether they constitute a new universality class with regard to $q$-Gaussian attractor distributions. We also study the generalized $q$-entropy production per unit time on the new unimodal dissipative maps, both for strong and weak chaotic cases. The $q$-sensitivity indices are obtained as well. Our results are, like those for the $z$-logistic maps, numerically compatible with the $q$-generalization of a Pesin-like identity for ensemble averages.
