12.07.2015 Views

Course in Probability Theory

Course in Probability Theory

Course in Probability Theory

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

1 CONDITIONING346. MARKOV PROPERTY . MARTINGALEfi at n . This example shows why <strong>in</strong> optional sampl<strong>in</strong>g the option may be takeneven with the knowledge of the present moment under certa<strong>in</strong> conditions . Inthe case here the present (namely fi A n) may leave one no choice!14 . In the gambler's ru<strong>in</strong> problem, suppose that S1 has the distributionP81 + (1 - P)8-1, P 0 iand let d = 2 p - 1 . Show that (_F(Sy) = d e (y) . Compute the probabilities ofru<strong>in</strong> by us<strong>in</strong>g difference equations to deduce (9(y), and vice versa .15 . Prove that for any L 1-bounded smart<strong>in</strong>gale {X,, ~i n , n E N,,, }, andany optional a, we have c (jXaI) < oo . [iurcr : Prove the result first for amart<strong>in</strong>gale, then use Doob's decomposition .]*16 . Let {Xn, n } be a mart<strong>in</strong>gale : x1 = X1, xn = Xn - X, -I for n > 2 ;let vn E :Tn_1 for n > 1 where A = 9,Tj ; now putnj=1vjxj .Show that IT,, 0~,7} is a mart<strong>in</strong>gale provided that Tn is <strong>in</strong>tegrable for every n .The mart<strong>in</strong>gale may be replaced by a smart<strong>in</strong>gale if v, > 0 for every n . Asa particular case take vn = l{n

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!