Existence of a Rate Function

Part of the Universitext book series (UTX)


Let E be a Polish space and π(x,•) a transition probability function on E. Set Ω = E η and let {Px: x ∈ E} be the Markov family on Ω with transition function π (i.e. Px (X(0) = x) = 1 and Px is a Markov process with transition function π). For n ≥ 1 and ω ∈ Ω, define
$$ {L_n}(\Gamma, \omega ) = \frac{1}{n}\sum\limits_0^{{n - 1}} {{X_{\Gamma }}(X(k,\omega ))\quad, \quad \Gamma \in {B_E}} $$
, and
$$ L_n^l(\Gamma, \omega ) = L_n^l(\Gamma, {\theta_1}\omega ) = \frac{1}{n}\quad \sum\limits_1^n {{X_{\Gamma }}(X(k,\omega ))} $$
10. Finally, define Qn,x and Qn,x on m1(m1E)) to be the distribution of Ln and L n 1 , respectively, under Px.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

Copyright information

© Springer-Verlag New York Inc. 1984

Authors and Affiliations

  1. 1.Department of MathematicsMassachusetts Institute of TechnologyCambridgeUSA

Personalised recommendations