Properties Of Random Walk Song Zhai Texas A&M University, Math Department Beihang University, Math Department 5/13/2015 Song Zhai Properties Of Random Walk What is random walk? Definition. A random walk is a path that is created by some stochastic process. Note. A random walk cares about not only the random phenomenon, but also what will happen as time or space changes. Song Zhai Properties Of Random Walk What is random walk? A simple but interesting example. Song Zhai Properties Of Random Walk What is random walk? Consider a person standing on the integer line who flips a coin and moves one unit to the right if it lands on heads, and one unit to the left if it lands on tails. The path that is created by the random movements of the walker is a random walk. Song Zhai Properties Of Random Walk What is random walk? Question. Suppose this person stands at zero point and he wants to go to x6=0 point. Can he get to x in finite steps? Or what is the probability that he can arrive at x in finite steps? Song Zhai Properties Of Random Walk Statement of probability space 1.(<,B,µ) B = Borel σ-field µ = Probability measure 2.×∞ i=1 < = {ω = (ω1 , ω2 , · · ·) : ωi ∈ <} 3.⊗∞ i=1 B = B ⊗ B ⊗ · · · 4.P = µ × µ × · · · ∞ Thus, our probability space will be (×∞ i=1 <, ⊗i=1 B, P). We can then define X1 , X2 , · · · i.i.d, taking values in ×∞ i=1 <, Xi (ω) = ωi . Song Zhai Properties Of Random Walk Stopping time I would like to introduce some knowledge of stopping time as my background. Definition. Fn = σ(X1 , X2 , · · ·, Xn ). Definition. A random variable N taking values in {1, 2, · · ·} ∪ {∞} is said to be a stopping time or an optional random variable if for every n< ∞, {N = n} ∈ Fn . Definition. Associated with each stopping time N is a σ-field FN = {A : A ∩ {N = n} ∈ Fn for all n < ∞}. Song Zhai Properties Of Random Walk Random walk Sn (with n fixed) Definition. Let X1 , X2 , · · · be i.i.d, taking values in ×∞ i=1 < and let Sn = X1 + X2 + · · · + Xn , Sn is a random walk. Song Zhai Properties Of Random Walk Random walk Sn (with n fixed) Definition. 1.A finite permutation π is a map from N onto N ,thus, (πω)i = ωπ(i) 2.An event A is permutable if π −1 A ≡ {ω : πω ∈ A} = A for any finite permutation π Song Zhai Properties Of Random Walk Random walk Sn (with n fixed) Hewitt-Savage 0-1 Law If X1 , X2 , · · · are i.i.d and A is permutable event, then P(A)∈ {0, 1} Song Zhai Properties Of Random Walk Random walk Sn (with n fixed) Theorem For a random walk on <, there are only four possibilities,one of which has probability one. (a)Sn =0 for all n (b)Sn → ∞, n → ∞ (c)Sn → −∞, n → ∞ (d)−∞ = liminfSn < limsupSn = ∞ Song Zhai Properties Of Random Walk Random walk Sn (with n fixed) Sketch of proof By checking that {ω : limsupSn (ω) = c} is a permutable event for all c∈ [−∞, ∞], we can apply Hewitt-Savage 0-1 law. For any fixed c0 , P({ω : limsupSn (ω) = c0 }) ∈ {0, 1}, which implies limsup Sn ≡ c. Song Zhai Properties Of Random Walk Random walk Sn (with n fixed) Let Sn0 = Sn+1 − X1 , since Sn0 has the same distribution as Sn , by taking limsup on both sides, we have c=c-X1 . Similarly, we have liminfSn ≡ b, and b=b-X1 . Song Zhai Properties Of Random Walk Random walk Sn (with n fixed) My results follows from some simple combinations. (1) c is finite. (2) b = −∞, c = −∞ (3) b = ∞, c = ∞ (4) b = −∞, c = ∞ Song Zhai Properties Of Random Walk Random walk SN (where N is stopping time) Before I talk about random walk SN , I would like to show you a fact. Fact. Let α(ω) = inf {n : ω1 + · · · + ωn > 0}, where inf = ∞, and set α(4) = ∞. then, we have, for arbitrary ω ∈ Ω, Sα (ω) = Sα(ω) (ω) > 0. Song Zhai Properties Of Random Walk Random walk SN (where N is stopping time) Notice that it’s not ture that for arbitrary ω ∈ Ω, Sn (ω) > 0. In that case, we say Sα (where α is stopping time) has some more beautiful properties than Sn . That’s the motivation why I want to talk about random walk SN (where N is stopping time). Song Zhai Properties Of Random Walk Random walk SN (where N is stopping time) Wald’s equation Let X1 , X2 , · · · be i.i.d with E |Xi | < ∞. If N is a stopping time with EN < ∞, then, ESN = EX1 EN. Song Zhai Properties Of Random Walk Random walk SN (where N is stopping time) Wald’s second equation Let X1 , X2 , · · · be i.i.d. with EX1 = 0 and EX12 = σ 2 < ∞. If T is a stopping time with ET < ∞, then, EST2 = σ 2 ET . Hint: First show that EST2 ∧n = E (T ∧ n)σ 2 . Then take the limit on both sides. Song Zhai Properties Of Random Walk Simple random walk A mathematical model of the question I gave at the beginning. Simple random walk Let X1 , X2 , · · · be i.i.d with P(Xi = 1) = 1 2 and P(Xi = −1) = 12 , let Tx = inf {n : Sn = x}, ∀x ∈ Z. (1) P(Tx < ∞) = 1 for all x ∈ Z, which means simple random walk can arrive at any integer in finite steps with probability 1. (2) ETx = ∞ for x 6= 0. Song Zhai Properties Of Random Walk Simple random walk Sketch of proof Let a < 0 < b be integers, and let N = inf {n : Sn ∈ / (a, b)}. By checking EN < ∞, we can apply Wald’s equation and ESN = EX1 EN = 0. Song Zhai Properties Of Random Walk Simple random walk By the definition, SN (ω)= a or b. So bP(SN = b) + aP(SN = a) = ESN = 0, since P(SN = b) + P(SN = a) = 1, solving this system of equation, we have ( P(SN = a) = P(SN = b) = Song Zhai b b−a −a b−a Properties Of Random Walk Simple random walk Let Ta = inf {n : Sn = a}, Tb = inf {n : Sn = b}, P(Ta < Tb ) = P(SN = a) = b b−a for all a < 0 < b Letting b → ∞ gives P(Ta < ∞) = 1 for all a < 0, because of the symmetry (and the fact T0 = 0), it follows that P(Tx < ∞) = 1 for all x ∈ Z, where Tx = inf {n : Sn = x}. Song Zhai Properties Of Random Walk Acknowledge Thanks Doctor Joel Zinn and Doctor Ursula Mueller-Harknett for helpful discussions. Song Zhai Properties Of Random Walk The end The End. Song Zhai Properties Of Random Walk