Scribd
Upload
192 views4 pages

Quiz-6 Answers and Solutions: Coursera. Stochastic Processes December 30, 2020

This document contains quiz questions and solutions about stochastic processes. It addresses whether certain processes are stationary, ergodic, or both. The key processes discussed are: 1) A process with Xt = cos(ωt + θ) which is shown to be ergodic and weakly stationary. 2) A process with Xt = εt + ξ cos(πt/12) which is not stationary but is ergodic. 3) A process with Yt = Xt+1 - Xt derived from Xt = α + βt + e-hλ, which is weakly stationary and ergodic. 4) The process Xt = e-

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
192 views4 pages

Quiz-6 Answers and Solutions: Coursera. Stochastic Processes December 30, 2020

This document contains quiz questions and solutions about stochastic processes. It addresses whether certain processes are stationary, ergodic, or both. The key processes discussed are: 1) A process with Xt = cos(ωt + θ) which is shown to be ergodic and weakly stationary. 2) A process with Xt = εt + ξ cos(πt/12) which is not stationary but is ergodic. 3) A process with Yt = Xt+1 - Xt derived from Xt = α + βt + e-hλ, which is weakly stationary and ergodic. 4) The process Xt = e-

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-6 answers and solutions

Coursera. Stochastic Processes


December 30, 2020

1. Let Xt = cos(ωt+θ) be a stochastic process and θ ∼ Unif[0, 2π], ω = π/10.


Is this process ergodic? Is it stationary?
Answer: It is ergodic and weakly stationary.
Solution:
Since the distribution of Xt is symmetric, its mean is 0.

K(t, s) = E(cos(wt + θ)cos(ws + θ))


1 1
= Ecos(w(t − s)) + Ecos(w(t + s) + 2θ)
2 2
= γ(t − s) + Ecos(w(t + s))Ecos(2θ) − Esin(w(t + s))Esin(2θ)
= γ(t − s),

because the means of cos(2θ) and sin(2θ) are equal to 0. Consequently,


this process is weakly stationary.
To prove that it is also ergodic, we need to look at the following:
T
1X 1 Tw
γ(r) = (cos0 + · · · + cos )
T r=0 2T 10
cw
≤ → 0,
2T
where cw depends on w. For instance, if w = 10, then cw ≥ 5, just cw = 5
to have a sharp bound.
Therefore, this process is ergodic.

2. Let Xt = εt +ξ cos(πt/12), t = 1, 2, ..., where ξ, ε1 , ε2 , ... are i.i.d. standard


normal random variables. Choose the correct statement.
Answer: Xt is not weakly stationary, but it is ergodic.
Solution:
The mean of the process is, obviously, nil, however, its covariance function
is equal to: K(t, s) = 1{t = s} Var ξt + cov(ξ cos(πt/12), ξ cos(πs/12)) =
1{t − s = 0} + cos(πt/12) cos(πs/12), which cannot be presented as a
function on (t − s). Thus, it is not stationary.
T
1X
E (εt + ξ cos(πt/12)) = 0
T t=0

1
T T
!
1X 1 X
Var (εt + ξ cos(πt/12)) = T+ cos2 (πt/12)
T t=0 T2 t=1
T
1 1 X
= + 2 cos2 (πt/12)
T T t=1
1 T 2
≤ + 2 = →0
T T T
as T → ∞.
1 PT
Therefore, the process is not stationary, but is ergodic, because E Xt →
T t=0
const.
3. Assume that for a process Xt it is known that E [Xt ] = α+βt, cov(Xt , Xt+h ) =
e−hλ for all h ≥ 0, t > 0, and some constants λ > 0, α, β. Is the process
Yt = Xt+1 − Xt stationary and ergodic?
Answer: Yt is weakly stationary and ergodic.
Solution:
E [Xt+1 − Xt ] = β does not depend on time. And, Yt = Xt+1 − Xt ,
clearly, has an autocovariance function. Hence, it is weakly stationary.
A strict stationarity is not the case, because, for instance, Y0 = X1 and
Y100 = X101 − X100 have different distribution laws:

Var X1 = 1
Var(X101 − X100 ) = 1 + 1 − 2 cov(X101 ; X100 ) = 2 − 2e−λ .

Additionally,
T
1X XT +1 − X0
(Yt ) = → const
T t=0 T
as T → ∞, so it is ergodic.
4. Let Xt = σWt + ct, where Wt is Brownian motion, σ, c > 0. Choose the
correct statements about this process:
Answer: Xt has continuous trajectories.

5. Consider the process Xn = Xn−1 +ξn , X0 = 0, ξn ∼ N (0, 1) ∀n ∈ N.


Choose the correct statements about this process.
T
 T 
1 P 1 P (T + 1)(2T + 1)
Answer: E Xt = 0, Var Xt =
T t=1 T t=1 6T
Solution:
Assuming T ≥ 3,

T
1X 1
E Xt = E[ξ1 + ξ1 + ξ2 + . . . ] = 0
T t=1 T

2
T
!
1X 1
Var Xt = Var(ξ1 + ξ1 + ξ2 + ξ1 + ξ2 + ξ3 + . . . )
T t=1 T2
1
= Var(T ξ1 + (T − 1)ξ2 + (T − 2)ξ3 + · · · + ξT )
T2
T
i.i.d. 1 2 2 1 X 2
= (T Var(ξ1 ) + (T − 1) Var(ξ 2 ) + . . . ) = t
T2 T 2 t=1
T (T + 1)(2T + 1) (T + 1)(2T + 1)
= 2
= → ∞,
6T 6T T →∞

⇒ the process is not ergodic.


6. Is it true that if two processes Xt and Yt have the same covariance func-
tions and Xt is ergodic, then Yt is also ergodic?
Answer: without any additional information — no
Solution: The process Xt is ergodic, if there exists such constant c that
T
1 P 1 PT
P
Xt → c as T → ∞, for which it is necessary that E Xt →
T t=1 T t=1 T →∞
 T 
1 P
const and Var Xt → 0. While from ergodicity of Xt the second
T t=1 T →∞
T
property is fulfilled for Yt , it is still possible that E T1
P
Yt → ∞: for
t=1
instance, take Xt = ξt and Yt = t + ξt , where ξi are i.i.d. random variables
for all i = 1, 2, . . ..

7. Let Xt = e−t We2t , where Wt is a Brownian motion. Find the mean and
the covariance function K(t, s) of this process.
Answer: EXt = 0, K(t, s) = e−|t−s|
Solution:
EXt = Ee−t We2t = e−t EWe2t = e−t · 0 = 0

K(t, s) = cov(e−t We2t , e−s We2s ) = e−(t+s) min{e2t , e2s }


(
e−t+s , s ≤ t,
=
e−s+t , s > t
= e−|t−s| .

8. Choose the correct statements about the process Xt = e−t We2t , where Wt
is a Brownian motion.
Answer: Xt is stationary, ergodic and continuous in the mean-squared
sense
Solution:
T
1X
EXt = 0, ⇒ E Xt = 0
T t=1

3
K(t, s) = cov(e−t We2t , e−s We2s ) = e−(t+s) min{e2t , e2s }
(
e−t+s , s ≤ t,
=
e−s+t , s > t
= e−|t−s| .

Thus, the process is weakly stationary, and since γ(t − s) = γ(u) =


e−|u| → 0, it is also ergodic. Now, as K(t, s) is continuous at (t0 , t0 ),
u→∞
Xt is continuous in the mean-squared sense, and since @γ 00 (0), it is not
stochastically differentiable.

You might also like