TAMS32/TEN1 STOKASTISKA PROCESSER TENTAMEN TORSDAG 25 AUGUSTI 2022 KL 14.00-18.00. Examinator och jourhavande lärare: Torkel Erhardsson, tel. 28 14 78. Permitted exam aids: Formel–och tabellsamling i TAMS32 Stokastiska processer (handed out during the exam). Mathematics Handbook for Science and Engineering (formerly BETA), by L. Råde och B. Westergren. Calculator with empty memories. The exam consists of 6 problems worth 3 points each. Grading limits : 8 points for grade 3, 11.5 points for grade 4, 15 points for grade 5. The results will be communicated by email. Problem 1 A stationary Gaussian process {X(t); t ∈ R}, with mean µX = 0 and spectral density SX (f ) = 1 ∀f ∈ R, is the input signal to a stable LTI with impulse response ( 2e−2t , for t ≥ 0; h(t) = 0, otherwise. Let {Y (t); t ∈ R} denote the output signal. (a) Compute the autocorrelation function of {Y (t); t ∈ R}. (b) Compute P (Y (1) > Y (0) + 1). Problem 2 Let {Xt ; t = 0, 1, . . .} be a Markov chain with state space SX = {1, 2, 3}, initial distribution p, and transition matrix P , where 0.25 0.75 0 0.2 P = 0.4 0.4 0.2 . p = 0.6 , 0.9 0 0.1 0.2 (a) Does the chain have a stationary and asymptotic distribution? The answer must be supported by an argument. If the answer is yes, compute this distribution. (b) Compute P (X3 = 1, X2 = 3|X1 = 2, X0 = 2). (c) Compute P (X3 = 1, X2 = 3|X1 6= 2, X0 = 2). Problem 3 Let X and Y be random variables such that X has the Exponential(1) distribution, while Y has the conditional pdf ( xe−xy , for y > 0; fY |X=x (y) = 0, otherwise. Compute the MMSE (the best predictor in the mean square sense) of X based on Y . Problem 4 Let {Yt ; t ∈ Z} be the wide sense stationary solution to the AR(1) equation Yt − aYt−1 = Xt ∀t ∈ Z, where {Xt ; t ∈ Z} is i.i.d. white noise with mean 0 and variance σ 2 < ∞, and 0 < |a| < 1. Since the noise is i.i.d., Xt is independent of (Y0 , . . . , Yt−1 ) for each t ∈ Z (you don’t have to show this). Define the random sequence {Ut ; t = 0, 1, . . .} by: Yt Ut = t ∀t = 0, 1, . . . a Is {Ut ; t = 0, 1, . . .} a martingale? The answer must be supported by an argument. Problem 5 A particle moves along the real line. It starts at a random position X1 , and has a random velocity X2 . The random variable (X1 , X2 ) has a two-dimensional Gaussian distribution, with the following mean vector and covariance matrix: 64 −8 10 . , CX = µX = −8 4 3 Let Y (t) = X1 + tX2 be the position of the particle at time t ≥ 0. (a) Compute the expectation function and the autocovariance function for the process {Y (t); t ≥ 0}. Is the process wide sense stationary? (b) Find a time t1 ≥ 0 such that Y (t1 ), the position of the particle at time t1 , is independent of the starting position Y (0). Problem 6 Let {X(t); t ≥ 0} be a Wiener process (Brownian motion) with variance parameter 0 < σ 2 < ∞. Let Z n X(t) dt ∀n = 1, 2, . . . Yn = t2 1 Does {Yn ; n = 1, 2, . . .} converge in mean square as n → ∞? Use a suitable criterion to prove or disprove.