MATH 5050/6815: Homework 3 Due on Tuesday, Mar. 6, by the end of lecture. For this homework set, you need the Jensen’s inqualities. The have to do with convex functions: A function f (x) is convex if for all 0 ≤ p ≤ 1, f (px + (1 − p)y) ≤ pf (x) + (1 − p)f (y). For more practice, please do Problem 5.3 page 125 (not for credit :-) ). Proposition 0.1 (Jensen’s Inequalities). Let f be a convex function so that the following expectations exist. For any random variable X and σ− algebra G, the following hold: (1) E(f (X)) ≥ f (E(X)), (2) E(f (X)|G) ≥ f (E(X|G)). Problem 1 a) Let Mn a mean µ positive martingale. Show that if the variance of the martingale V ar(Mn ) < C for al n then the martingale is uniformly integrable. b) Problem 5.4 page 126 from the book. Problem 2: Branching processes Let Xn be the population size of the n-th generation of a branching process with X0 = 1, with offspring distribution Z, with E(Z) = µ > 1 and V ar(Z) = σ 2 < ∞. For notation purposes, assume that the i-th individual of generation n PXn−1 (i) (i) gave birth to Zn many children and so Xn = i=1 Zn−1 . 2 2 (1) Find the E(Xn ) in terms of E(Xn−1 ). (Hint: Condition on Xn−1 ) (2) Find the E(Xn2 ) by solving the recursion you found in the previous question. (3) Find the V ar(Xn ). (4) Show that the martingale Mn = Xn /µn is uniformly integrable. (5) Show that the probability of the branching process surviving is positive. Problem 3 Let X a F-measurable random variable and {Fn }n≥0 a filtration of σ− algebras, so that Fn ⊆ F for all n ∈ N. Also assume E|X| < ∞ and EX = µ, V ar(X) = σ 2 < ∞. (1) Let Mn = E(X|Fn ). Show that Mn is a mean µ, Fn − martingale. (2) Decide whether Mn is uniformly integrable. S (3) Explain why you can apply martingale convergence theorem. Suppose that n≥0 Fn = F. Can you guess what M∞ should be? Why? Problem 4 Do problem 5.17, page 129 in your book. Problem 5 10 points, Extra Credit: Problem 5.6. , page 126 from your book. 1