Scribd
Upload
120 views4 pages

Quiz-5 Answers and Solutions: Coursera. Stochastic Processes December 30, 2020

This document contains solutions to 7 questions about stochastic processes from a Coursera quiz. Key results include: 1) The covariance function of the process Xt = Wt+h - Wt is K(t,s) = h - |t - s| for |t - s| <= h and 0 otherwise. 2) If the increments of a process Xt are stationary, then Xt is stationary. 3) For the process Yn+1 = αYn + Xn, Yn is not stationary but has mean 0 and variance (1 - α2n)/(1 - α2). 4) The process Xt = cosξt + sin

Uploaded by

sirj0_hn
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
120 views4 pages

Quiz-5 Answers and Solutions: Coursera. Stochastic Processes December 30, 2020

This document contains solutions to 7 questions about stochastic processes from a Coursera quiz. Key results include: 1) The covariance function of the process Xt = Wt+h - Wt is K(t,s) = h - |t - s| for |t - s| <= h and 0 otherwise. 2) If the increments of a process Xt are stationary, then Xt is stationary. 3) For the process Yn+1 = αYn + Xn, Yn is not stationary but has mean 0 and variance (1 - α2n)/(1 - α2). 4) The process Xt = cosξt + sin

Uploaded by

sirj0_hn
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 4

Quiz-5 answers and solutions

Coursera. Stochastic Processes


December 30, 2020

1. Let Wt be a Brownian Motion and h > 0 be a fixed number. Find a


covariance function of the process Xt = Wt+h − Wt .
(
h − |t − s|, if |t − s| ≤ h
Answer: K(t, s) =
0, if |t − s| > h
Solution: K(t, s) = cov(Wt+h − Wt , Ws+h − Ws )

(
0, if t>s+h
=
cov(Wt+h − Ws+h + Ws+h − Wt , Ws+h − Ws ), if t≤s+h

(
0, if t>s+h
=
cov(Ws+h − Wt , Ws+h − Wt + Wt − Ws ), if t≤s+h
(
0, if t>s+h
=
Var(Ws+h − Wt ), if t≤s+h
(
0, if t>s+h
=
s + h − t, if t≤s+h

2. Let Xt is a process with independent and stationary increments and h is a


positive constant. Moreover, EXt = 0 and EXt2 < ∞. Is Yt = Xt+h − Xt
a wide-sense stationary process ?
Answer: Yes
Hint: If increments of the process Xt are stationary, then Xt is also
stationary.
3. Let Yn be a stochastic process which is defined as follows: Yn+1 = αYn +
Xn , n = 0, 1, ... Assume Y0 = 0, |α| < 1 and Xn is a sequence of i.i.d.
standard normal random variables for n = 0, 1, 2, ... Determine whether
Yn is stationary and find its mean and variance:
1 − α2n
Answer: Yn is non-stationary, EYn = 0, V arYn =
1 − α2
Solution: The key for solution lies in the finding the covariance function:

1
K(t, s) = Cov(Yt ; Ys )
= Cov(αt−1 X0 + · · · + α0 Xt−1 ; αs−1 X0 + · · · + α0 Xs−1 )
= αt−1 αs−1 + αt−2 αs−2 + · · · + αt−s+1 α + αt−s α
= αt−s (α2s−2 + α2s−4 + · · · + 1)
1 − α2s
= αt−s .
1 − α2

4. Consider the process Xt = cos ξt + sin ξt−1 , where ξt ∼ U nif ([0, 2π]) for
all t ∈ N. Choose the correct statements about this process.
Answer: EXt = 0 and Xt is weakly stationary
Solution:

E[cos ξt + sin ξt−1 ] = E cos ξt + E sin ξt−1 = 0

K(t, s) = cov(cos ξt + sin ξt−1 , cos ξs + sin ξs−1 )


= E[cos ξt cos ξs ] + E[cos ξt sin ξs−1 ]
+E[sin ξt−1 cos ξs ] + E[sin ξt−1 sin ξs−1 ]
2 2
= (E cos ξt + E sin ξt−1 )I{t = s} + (E cos ξt sin ξs−1 )I{t − s = −1}
+(E sin ξt−1 cos ξs )I{t − s = 1}.

Since
1
E sin ξt cos ξt = E[sin 2ξt ] = 0
2
and E cos2 ξt + E sin2 ξt−1 = 1, we get that K(t, s) = 1 · I{t = s}, and the
process is weakly stationary.
5. Consider the process

Xt = ξt+2 − ξt+1 + ξt , ξ ∼ i. i. d. N (0, 1) for all t = 0, 1, 2, . . . .

Find the mean and the covariance function of this process and determine
whether it is weakly stationary.
Answer: Xt is weakly stationary.

E[Xt ] = 0,


3, t = s,

−2, |t − s| = 1,
K(t, s) =


1, |t − s| = 2,
0, |t − s| > 2.

Solution:

E[Xt ] = E[ξt+2 − ξt+1 + ξt ] = E[ξt+2 ] − E[ξt+1 ] + E[ξt ] = 0,

2
cov(Xt , Xs ) = cov(ξt+2 − ξt+1 + ξt , ξs+2 − ξs+1 + ξs )
= (1 + 1 + 1)I{t = s} + (−1 − 1)I{|t − s| = 1} + I{|t − s| = 2}


 3, t = s,

−2, |t − s| = 1,
=


 1, |t − s| = 2,
0, |t − s| > 2.

Since E[Xt ] = const and K(t, s) depends only on the difference t − s, the
process is weakly stationary.

6. Consider the process

Xt = ξt+2 − ξt+1 + ξt , ξt ∼ i. i. d. N (0, 1) for all t = 0, 1, 2, . . . .

Find the spectral density of this process.


Answer:
1
g(u) = (3 − 4 cos u + 2 cos 2u).
2π
Solution: Since Xt is weakly stationary and has the autocovariance func-
tion 

 3, x = 0,

−2, |x| = 1,
γ(x) =


 1, |x| = 2,
0, |x| > 2,

the spectral density g(u) can be calculated as


1 1
3 − 2eiu − 2e−iu + e2iu + e−2iu =

g(u) = (3 − 4 cos u + 2 cos 2u) .
2π 2π

7. Find the spectral density of the process

Yt = Xt + Xt−1 + Xt−2 ,

where

Xt = ξt+2 − ξt+1 + ξt , ξt ∼ i. i. d. N (0, 1) for all t = 0, 1, 2, . . . .

1

Answer: gY (u) = 2π 9 + 12 cos 2u − 16 cos2 u + 4 cos2 2u .
Solution: Since Yt is the linear filter for the process Xt , its spectral
density can be calculated as

gY (u) = gX (u)|F[ρ](u)|2 ,

where ρ(h) = 1{h = 0} + 1{h = 1} + 1{h = 2} and gX (u) is the spectral


density of Xt . Thus,

F[ρ](u) = 1 + eiu + e2iu

3
and

|F[ρ](u)|2 = F F̄(u)
= (1 + eiu + e2iu )(1 + e−iu + e−2iu ) = 3 + 2(eiu + e−iu ) + (e2iu + e−2iu )
= 3 + 4 cos u + 2 cos 2u

and
1
gY (u) = (3 − 4 cos u + 2 cos 2u) (3 + 4 cos u + 2 cos 2u)
2π
1
9 + 12 cos 2u − 16 cos2 u + 4 cos2 2u .

=
2π

You might also like