I Stochastic process.pptVIP

* * * * * * * * * Lemma 1.4.2 Let (?, F, P ) be a probability space , P(B)0, Let ? be a random variable ,The expectation,if it exists, of ? on probability space (?, F, PB ) is called conditional expectation given B and is denoted by E(?|B). Remark: let ?= IA, then * Conditional expectation given C Constructive definition :Let {Bn,n=1,2,...} be a countable partition of ? and let C= ? {Bn,n=1,2,...} , let ? be the family of all random variable’s ? whose expectation E(?) exists.consider the elementary functions E(? | C)=∑[E(?| Bn )] IBn ? ? ? is called conditional expectation of ? given C. * Example1.4.3 Toss a coin 2 times, The collection of all possible outcomes is ?= {HH; HT; TH;TT }. collection F1 of sets determined by the first toss consists of: H= {HH; HT},T= {TH;TT},that is F1 =?{H,T}. Let X be a random variable defined on ?: X(HH)=3,X(HT)=X(TH)=2,X(TT)=1. E(X| F1)=? * Theorem 1.4.4 E(? | C) is C.-measurable,and Descriptive definition Definition 1.4.5 The conditional expectation E(? | C) is the function from ? to Rn satisfying : E(? | C) is C measurable. (2) * Appendex :Radon-Nikodym theorem Let ? and ? are finite measures on (?, F), ? is said to be absolutely continuous with respect to ?, if ?(A)=0 implies ?(A)=0 .This is denoted by writing ? ?. Th (Radon-Nikodym) Let ? be a finite measure in a measure space (?, F) and ? be another finite measure that is absolutely continuous with respect to ? . (ie ? ?). Then there is a “unique” non negative function X ?L1(?, F ,?) such that Denot by * Properties Suppose E? and E(? |C) exists. E(? | C) = E(?+ | C) - E(?- | C) a.s. If ? is C-measurable,then E(? | C) = ? a.s. If ? =a?R, a.s. ,then E(? | C) =a a.s. If ? is independent to C, then E(? |C) =E ? a.s. If ? is C-measurable and E?? exists,then E(?? | C)= ?E(? | C) a.s. * (6)Let C ? C1 ?F, then E(E(? | C)| C1 )= E(? | C) a.s. If E(? | C1 ) exists then E(E(? | C1 ) | C )= E(? | C) a.s (7) Let ? is a measura