EE 325: Probability and Random Processes

Final Exam : 3 hrs

1.
(2x + y)
f (x, y) = 0 < x < 1, 0 < y < 2 (1)
4
= 0 elsewhere (2)

Find the marginal fX (x) and fY |X (y|x = 14 ).

Sol: Z
(2x + y)
2 1
fX (x) = dy = x + , 0 < x < 1
0 4 2
f (x, y) (2x + y)/4 (1/2 + y)
f (y|x) = = = , 0 < y < 2.
f (x) x + 1/2 3
2. X1 and X2 be two continuous random variables having the joint probability density.

f (x1 , x2 ) = 1 0 < x1 , x2 < 1 (3)

= 0 elsewhere (4)

Find the joint probability density of Y1 = X1 +X2 and Y2 = X2 . Find marginal density
of Y1 .
 
 1 1 
J =  =1

0 1 

f (y1 , y2 ) = 1, 0 < y2 < 1, y2 < y1 < 1 + y2

f (y1 ) = y1 , 0 < y1 ≤ 1 (5)

= 1 − y1 , 1 < y1 < 2 (6)

1
 y2

1 2
y1
qP
n
3. Let Y = Xi2 , where Xi are IID, N (0, 1). Find probability density of Y .
i=1
√
Sol: For Z = ni=1 Xi2 , we have fZ (z) = Γ(n/2, 2). Y = Z
P

fY (y) = 2yfZ (y 2 ), y > 0

where n
z 2 −1 e−z/2
FZ (z) = n .
2 2 Γ(n/2)
4. Phone calls to a telephone exchange are modelled as a poisson process with rate λ.
Given IID observations t1 , . . . , tn of the first arrival time T , find the MLE estimate of
λ. Is it unbiased ? Find probability density of the MLE estimator.
Sol: n n
n
X X
f (t, . . . , tn ) = λ exp(−λ ti ), ln f = n ln λ − λ ti .
i=1 i=1

Differentiating

n
n X n
− ti = 0 → λ̂ = Pn
λ i=1 i=1 ti
Pn
for z = i=1 ti , we have fz (z) = Γ(n, λ−1 ). Then

Z ∞ n z n−1 λn e−λz nλ
E(n/z) = = .
0 z (n − 1)! n−1
Its a biased estimator.
λ̂ = n/z, therefore its density is

n n
f (λ) = fz ( ), λ > 0
λ 2 λ

2
5. Given a inhomogeneous Poisson process with rate λ(t), find the probability density of
first arrival time.
Sol: Z t
fT (t) = λ(t) exp(− λ(τ )dτ ).
0

6. Let W (t) be a Brownian motion with parameter α and W (0) = 0. Find the probability
density of maximum value of W over the interval [0, T ].
Sol: Let Z denote the maximum value. For a > 0,

FZ (a) = P r(T1 > T ) = 1 − FT1 (T ),

where T1 is the first passage time for crossing a.

1 1
P r(W (T ) < a) = FT1 (T ) + (1 − FT1 (T )) = 1 − FT1 (T )
2 2
Ra w2
FT1 (T ) = 2(1 − P r(W (T ) < a)), where P r(W (T ) < a) = √ 1 exp(− 2αT )dw.
−∞ 2παT

7. Given a linear system with impulse response h(t) = exp(− Tt ), t > 0, with white
noise input X(t) such that RXX (τ ) = N20 δ(τ ), find the correlation function RY Y (τ )
and power spectral density SY Y (ω) of output Y (t).
1
Sol:H(ω) = 1
+jω
, therefore SY Y (ω) = |H(ω)|2 SXX (ω).
T

N0 1
SY Y (ω) = 1
2 T2
+ ω2

taking inverse Fourier transform

N0 T |τ |
SY Y (τ ) = exp(− )
4 T
8. Write the state transition matrix P of the Markov chain below. Find the fundamental
matrix N . Find the average number of times you visit state 3, starting from 2, before
absorbtion. Find the probability of absorbtion in state 1, if you start in state 2.
Sol:
2 0 2/3 1/3 0
 

3  1/3 0 0 2/3 
P = 
.

1 0 0 1 0 


4 0 0 0 1

3
 1/3 2/3

1
1
4
1 2 1/3 3 2/3

" # " #
0 2/3 1/3 0
Q= , R=
1/3 0 0 2/3

N = (I − Q)−1 , implying
" # " #
9/7 6/7 3/7 4/7
N= , NR =
3/7 9/7 1/7 6/7

Starting from 2, you visit state 3, on average 6/7 number of times (N12 ). Probability
of absorbtion in 1, when you start in 2 is 3/7 (NR11 ).