PAPER CODE NO. MATH 360 EXAMINER: Dr. Yi Zhang, TEL.NO. 44761 DEPARTMENT: Mathematical Sciences MAY 2019 EXAMINATIONS Applied Stochastic Models Time allowed: Two and a half hours INSTRUCTIONS TO CANDIDATES: Full marks will be given for complete answers to FIVE questions. Only the best FIVE will be taken into account. Paper Code MATH 360 Page 1 of 6 CONTINUED 1. (a) Consider a discrete-time Markov chain with the state space {1, 2, 3} and the (one-step) transition probability matrix P given by P = 0 1 00 1 2 1 2 1 2 0 1 2 . Compute p (n) 11 for each n ≥ 0. [13 marks] (b) A system can be in one of the states 1, 2, 3 and 4. If the system is at state j with j < 4, then, on the next step, it passes to state j + 1. From state 4, independently on everything else, on the next step, the system passes either to 2 or to 3 with equal probabilities p˜42 = p˜43 = 1 2 . (i) Model this system as a discrete-time Markov chain with the state space E and the (one-step) transition matrix say P˜ = {p˜ij}i,j∈E, and compute P˜ 2. (You need explicitly state what E and P˜ are.) [3 marks] (ii) Compute p˜ (16) 22 . [4 marks] 2. A machine produces two items every day. The probability that an item is non-defective is p ∈ (0, 1) with p 6= 1 2 (all items are independent), and defective items are thrown away instantly. The demand is one item per day, and any demand that cannot be satisfied by the end of the day is lost, while any extra item is stored. Let Xn be the number of items in storage just before the beginning of the nth day. (a) Model {Xn}n≥0 as a discrete-time Markov chain, that is, indicate its state space and transition probability matrix. [4 marks] (b) Decide for what values of p ∈ (0, 1) with p 6= 1 2 , {Xn}n≥0 is an ergodic discrete-time Markov chain, and obtain the corresponding invariant distribution. [8 marks] (c) Decide for what values of p ∈ (0, 1), {Xn}n≥0 is a null recurrent discrete- time Markov chain. [8 marks] Paper Code MATH 360 Page 2 of 6 CONTINUED 3. (a) Consider a continuous-time Markov chain with the state space {1, 2} and the Q-matrix given by Q = ( −α α β −β ) , where α, β > 0. Compute p12(t) and p11(t) for each t ≥ 0. [12 marks] (b) Consider {Xt}t≥0 and {Yt}t≥0 as two independent continuous-time Markov chains on the common state space {1, 2} with the common Q-matrix Q = ( −α α β −β ) , where α, β > 0. Decide with justifications whether {(Xt, Yt)}t≥0 is a continuous- time Markov chain. If it is, indicate its state space and Q-matrix. [5 marks] (c) Consider a continuous-time Markov chain with the state space {a, b, c, d} and the Q-matrix given by Q = −2 1 0 1 1 −2 1 0 0 1 −2 1 1 0 1 −2 Compute pa,a(t) for all t ≥ 0. [3 marks] Paper Code MATH 360 Page 3 of 6 CONTINUED 4. (a) Consider a birth-and-death process with birth rate λn > 0 for all n = 0, 1, . . . , and death rate µn for all n ≥ 1. Consider the series ∞∑ n=1 ( 1 λn + µn λnλn−1 + · · ·+ µnµn−1 . . . µ1 λnλn−1 . . . λ1λ0 ) . State without proof the connection between the regularity of the birth-and-death process with the convergence or divergence of the above series. [2 marks] (b) For each of the following birth-and-death processes, decide with reasons whether it is regular: (i) λn = n+ 2, n ≥ 0; µn = n, n ≥ 1. (ii) λn = n+ 1, n ≥ 0; µn = n, n ≥ 1. (iii) λ0 = λ1 = 1, λn = n− 1, n ≥ 2; µn = n, n ≥ 1. [9 marks] (c) The following fact is known. For a discrete-time Markov chain on E = {0, 1, . . . } with the transition probability given by p0,1 = p0 = 1; pi,i+1 = pi ∈ (0, 1), pi,i−1 = qi = 1− pi, ∀ i ≥ 1, it is recurrent if and only if ∞∑ k=1 qkqk−1 . . . q1 pkpk−1 . . . p1 =∞. For each of the three birth-and-death processes in part (b), decide whether it is recurrent. [9 marks] Paper Code MATH 360 Page 4 of 6 CONTINUED 5. Let {Xt}t≥0 be a continuous-time Markov chain taking values in a finite space {1, 2, 3, 4} with the Q-matrix given by Q = −3 1 1 1 0 −3 2 1 1 2 −4 1 0 0 1 −1 . Suppose P (X0 = 1) = 1. Calculate the expected time duration until the process reaches state 2 for the first time, that is, the expected hitting time to state 2 given the initial state 1. [7 marks] (b) Jobs arrive according to a Poisson process of rate λ > 0. They get processed individually first-in-first-served, by a single processor, the processing times being independent random variables, each with exponential distribution of parameter ν > 0. After processing, a job either leaves the system with probability p (0 < p < 1), or with probability 1 − p it is split into two separate jobs, which are both sent to join the queue for processing again. Let Xt be the number of jobs in the system at time t ≥ 0. (i) Model {Xt}t≥0 as a continuous-time Markov chain, that is, indicate its state space E and Q-matrix. Decide when, in terms of the values of p, ν, λ, this continuous-time Markov chain is ergodic, and obtain limt→∞ P (Xt = j) for each j ≥ 1. [10 marks] (ii) Compute the expected return time to state 0, given the initial state 0. [3 marks] Paper Code MATH 360 Page 5 of 6 CONTINUED 6. (a) Let {Zt}t≥0 be a (standard) Brownian motion. Check, with justifi- cations, whether or not each of the following processes is a standard Brownian motion. (i) Ut = Z 2 t ; (ii) Vt = Zt + Zt2 ; (iii) Wt = Zt+2 − Z2. [9 marks] (b) Consider a population whose members cannot die, and each member is independent of the others, and gives birth to descendants according to a Poisson process with rate λ > 0. Let {Xt}t≥0 denote the population process. Assume that the initial population is not empty. So we take the state space E = {1, 2, . . . }. Show that the transition probability function of {Xt}t≥0 is given by pij(t) = (j − 1)! (j − i)!(i− 1)! e−iλt(1− e−λt)j−i for each j ≥ i ≥ 1, and pij(t) = 0 if j < i. (Here you may use the fact that∑ j∈E pij(t) = 1 for each i ≥ 1 and t ≥ 0.) [7 marks] (c) Let {Zt}t≥0 be a (standard) Brownian motion. Decide for what values of θ ∈ R = (−∞,∞), Xt = e θZt+ θ 2 t 2 defines a martingale with respect to the history of the standard Brownian motion {Zt}. [4 marks] 7. Let {Wt}t≥0 be a standard Brownian motion. (a) Solve the stochastic differential equation dXt = Xtdt+ dWt with some initial condition X0 = Z. [16 marks] (b) Obtain the stochastic differential equation satisfied by {Xt}t≥0 with Xt = Wt 1 + t . [4 marks] Paper Code MATH 360 Page 6 of 6 END
欢迎咨询51作业君