On estimating the memory for finitarily Markovian processes

نویسندگان

  • Gusztáv Morvai
  • Benjamin Weiss
چکیده

Finitarily Markovian processes are those processes {Xn}n=−∞ for which there is a finite K (K = K({Xn} 0 n=−∞) such that the conditional distribution of X1 given the entire past is equal to the conditional distribution of X1 given only {Xn} 0 n=1−K . The least such value of K is called the memory length. We give a rather complete analysis of the problems of universally estimating the least such value of K, both in the backward sense that we have just described and in the forward sense, where one observes successive values of {Xn} for n ≥ 0 and asks for the least value K such that the conditional distribution of Xn+1 given {Xi} n i=n−K+1 is the same as the conditional distribution of Xn+1 given {Xi} n i=−∞. We allow for finite or countably infinite alphabet size. Les processus Markoviens finitaires sont des processus {Xn} ∞ n=−∞ pour lesquels il existe un entier K fini (K = K({Xn} 0 n=−∞) tel que la distribution conditionnelle de X1 etant donné tout le passé soit égale à la distribution conditionnelle de X1 etant donné seulement {Xn} 0 n=1−K . La plus petite valeur d’un tel K est appelée la longueur de la mémoire. Nous donnons une analyse complète du problème de l’estimation de la plus petite de ces valeurs de K, aussi bien en remontant dans le passé qu’en allant vers le futur, c’est à dire quand on observe les valeurs successives de {Xn} pour n ≥ 0 et qu’on recherche la plus petite valeur de K telle que la distribution conditionnelle de Xn+1 etant donné {Xi} n i=n−K+1 soit la même que la distribution conditionnelle de Xn+1 etant donné {Xi} n i=−∞. La taille des alphabets peut etre choisie finie ou infinie.

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

A note on prediction for discrete time series

Let {Xn} be a stationary and ergodic time series taking values from a finite or countably infinite set X and that f(X) is a function of the process with finite second moment. Assume that the distribution of the process is otherwise unknown. We construct a sequence of stopping times λn along which we will be able to estimate the conditional expectation E(f(Xλn+1)|X0, . . . , Xλn) from the observ...

متن کامل

Prediction for discrete time series

Let {Xn} be a stationary and ergodic time series taking values from a finite or countably infinite set X . Assume that the distribution of the process is otherwise unknown. We propose a sequence of stopping times λn along which we will be able to estimate the conditional probability P (Xλn+1 = x|X0, . . . , Xλn) from data segment (X0, . . . , Xλn) in a pointwise consistent way for a restricted ...

متن کامل

Finite Factors of Bernoulli Schemes and Distinguishing Labelings of Directed Graphs

A labeling of a graph is a function from the vertices of the graph to some finite set. In 1996, Albertson and Collins defined distinguishing labelings of undirected graphs. Their definition easily extends to directed graphs. Let G be a directed graph associated to the k-block presentation of a Bernoulli scheme X . We determine the automorphism group of G, and thus the distinguishing labelings o...

متن کامل

Simulating memory effects with discrete dynamical systems

We propose using discrete dynamical systems to model non-Markovian processes. This implies a whole hierarchy of dynamical systems with dimensionality increasing proportional to the memory. Specific long-range memory effects are investigated for a quadratic, a cubic and a quartic map. Markovian processes transform the quadratic map into a normal form of the logistic equation; the H6non map corre...

متن کامل

Quantum semi-Markov processes.

We construct a large class of non-Markovian master equations that describe the dynamics of open quantum systems featuring strong memory effects, which relies on a quantum generalization of the concept of classical semi-Markov processes. General conditions for the complete positivity of the corresponding quantum dynamical maps are formulated. The resulting non-Markovian quantum processes allow t...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:
  • CoRR

دوره abs/0712.0105  شماره 

صفحات  -

تاریخ انتشار 2007