Math 546 Assignment 2 (due Tues. Oct. 20) 1. Assume S and T are (Ft0 )-stopping times and Z is an FS0 -measurable random variable. Show that Z1{S≤T } is FT0 -measurable. 2. (a) Assume T is an (Ft0 )-stopping time and A ∈ FT0 . Show that TA (ω) = T (ω) if ω ∈ A; ∞ otherwise, is also an (Ft0 )-stopping time. 0 (b) Assume {Xt : 0 ≤ t ≤ ∞} is a cadlag (Ft0 )-adapted process, where F∞ = F. If for 0 every (Ft )-stopping time T , XT is an integrable r.v. such that E(XT ) = E(X0 ), show that X is a u.i. (Ft0 )-martingale. 3. If B is a standard one-dimensional Brownian motion, and√Ta = inf{t ≥ 0 : |Bt | = a} for a ≥ 0, prove that for all λ ≥ 0, E(exp(−λTa )) = sech( 2λ a), where sech x is the hyperbolic secant function. Do this by first showing that M (t) = E(t ∧ Ta ) is a u.i. (FtB )-martingale, where E(t) = exp(αBt − α2 t/2). 4. Assume {Xt : t ≥ 0} is a continuous (Ft0 )-adapted process taking values in Rd and K is a compact subset of Rd . Prove that DK (ω) = inf{t ≥ 0 : X(t, ω) ∈ K} is an (Ft0 )-stopping time. Do this directly without using the Projection Theorem. Hint: K = ∩n K 1/n , where K 1/n is the open set of points which are less than 1/n away from K. Now write {DK ≤ t} as an event involving rational times and K 1/n . 5. Law of the iterated logarithm. The goal of this exercise is to prove that for a standard one-dimensional Brownian motion B, lim sup t→0+ Bt = 1 a.s., φ(t) p where φ(t) = 2t log log(1/t) for t < e−1 . Let Mt = sups≤t Bs . Step 1. Let r and c be real numbers such that 1 < 1/r < c2 . Bound the probabilities P (Mrn−1 > cφ(rn )) when n → ∞ and infer that a.s., lim sup t→0+ Bt ≤ 1 a.s. φ(t) Step 2. Show that for each 0 < r < 1 w.p. 1 there are infinitely many n so that Brn − Brn+1 ≥ √ 1 − r φ(rn ). Conclude that the statement at the beginning of the exercise holds.