Computer Science > Information Theory
[Submitted on 8 Jul 2012 (v1), last revised 22 Aug 2013 (this version, v2)]
Title:Zipf and non-Zipf Laws for Homogeneous Markov Chain
View PDFAbstract:Let us consider a homogeneous Markov chain with discrete time and with a finite set of states $E_0,\ldots,E_n$ such that the state $E_0$ is absorbing, states $E_1,\ldots,E_n$ are nonrecurrent. The goal of this work is to study frequencies of trajectories in this chain, i.e., "words" composed of symbols $E_1,\ldots,E_n$ ending with the "space" $E_0$.
Let us order words according to their probabilities; denote by $p(t)$ the probability of the $t$th word in this list. In this paper we prove that in a typical case the asymptotics of the function $p(t)$ has a power character, and define its exponent from the matrix of transition probabilities. If this matrix is block-diagonal, then with some specific values of transition probabilities the power asymptotics gets (logarithmic) addends. But if this matrix is rather sparse, then probabilities quickly decrease; namely, the rate of asymptotics is greater than that of the power one, but not greater than that of the exponential one. We also establish necessary and sufficient conditions for the exponential order of decrease and obtain a formula for determining the exponent from the transition probability matrix and the initial distribution vector.
Submission history
From: Eduard Lerner [view email][v1] Sun, 8 Jul 2012 13:50:02 UTC (77 KB)
[v2] Thu, 22 Aug 2013 04:41:17 UTC (148 KB)
Current browse context:
cs.IT
References & Citations
Bibliographic and Citation Tools
Bibliographic Explorer (What is the Explorer?)
Connected Papers (What is Connected Papers?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)
Code, Data and Media Associated with this Article
alphaXiv (What is alphaXiv?)
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Hugging Face (What is Huggingface?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)
Demos
Recommenders and Search Tools
Influence Flower (What are Influence Flowers?)
CORE Recommender (What is CORE?)
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.