Probability and Stochastic Processes (ENCS 6161)

SOLUTIONS – ASSIGNMENT NO. 5

Problem No. 6.7

𝑓𝑋𝑌𝑍 (𝑥, 𝑦, 𝑧) = 𝑘(𝑥 + 𝑦 + 𝑧), 0 ≤ 𝑥 ≤ 1, 0 ≤ 𝑦 ≤ 1, 0 ≤ 𝑧 ≤ 1
(a)
1 1 1
1 = ∫ ∫ ∫ 𝑘(𝑥 + 𝑦 + 𝑧)𝑑𝑥 𝑑𝑦 𝑑𝑧
0 0 0
1 1 1
𝑥2
1 = 𝑘 ∫ ∫ [ + 𝑥𝑦 + 𝑥𝑧] 𝑑𝑦 𝑑𝑧
0 0 2 0
1 1
1
1 = 𝑘 ∫ ∫ ( + 𝑦 + 𝑧) 𝑑𝑦 𝑑𝑧
0 0 2
1 1
1 𝑦2
1=𝑘 ∫ [ 𝑦+ + 𝑧𝑦] 𝑑𝑧
0 2 2 0
1
1 1
1 = 𝑘 ∫ ( + + 𝑧) 𝑑𝑧
0 2 2
1
1 1 𝑧2
1 = 𝑘[ 𝑧+ 𝑧+ ]
2 2 2 0
3
1 = 𝑘( )
2
2
𝑘=
3

Page 1 of 8
(b)
𝑓𝑋𝑌𝑍 (𝑥, 𝑦, 𝑧)
𝑓𝑋 (𝑥|𝑦, 𝑧) =
𝑓𝑌𝑍 (𝑦, 𝑧)
𝑓𝑋𝑌𝑍 (𝑥, 𝑦, 𝑧)
𝑓𝑍 (𝑧|𝑥, 𝑦) =
𝑓𝑋𝑌 (𝑥, 𝑦)

1
2 1 2 𝑧2 2 1
𝑓𝑋𝑌 (𝑥, 𝑦) = ∫ (𝑥 + 𝑦 + 𝑧)𝑑𝑧 = [𝑥𝑧 + 𝑦𝑧 + ] = (𝑥 + 𝑦 + )
3 0 3 2 0 3 2
1
2 1 2 𝑥2 2 1
𝑓𝑌𝑍 (𝑦, 𝑧) = ∫ (𝑥 + 𝑦 + 𝑧)𝑑𝑥 = [ + 𝑥𝑦 + 𝑥𝑧] = ( + 𝑦 + 𝑧)
3 0 3 2 0
3 2
𝑓𝑋𝑌𝑍 (𝑥, 𝑦, 𝑧) (𝑥 + 𝑦 + 𝑧)
𝑓𝑋 (𝑥|𝑦, 𝑧) = = , 0 ≤ 𝑥 ≤ 1, 0 ≤ 𝑦 ≤ 𝑧, 0 ≤ 𝑧 ≤ 1
𝑓𝑌𝑍 (𝑦, 𝑧) 1
2+𝑦+𝑧
𝑓𝑋𝑌𝑍 (𝑥, 𝑦, 𝑧) (𝑥 + 𝑦 + 𝑧)
𝑓𝑍 (𝑧|𝑥, 𝑦) = = , 0 ≤ 𝑥 ≤ 1, 0 ≤ 𝑦 ≤ 1, 0 ≤ 𝑧 ≤ 1
𝑓𝑋𝑌 (𝑥, 𝑦) 1
𝑥+𝑦+2

(c)
1
2 1 1
𝑓𝑋 (𝑥) = ∫ 𝑓𝑋𝑌 (𝑥, 𝑦)𝑑𝑦 = ∫ (𝑥 + 𝑦 + ) 𝑑𝑦
0 3 0 2
1
2 𝑦2 1 2
𝑓𝑋 (𝑥) = [𝑥𝑦 + + 𝑦] = (𝑥 + 1), 0≤𝑥≤1
3 2 2 0 3

1
2 1 1
𝑓𝑌 (𝑦) = ∫ 𝑓𝑋𝑌 (𝑥, 𝑦)𝑑𝑥 = ∫ (𝑥 + 𝑦 + ) 𝑑𝑥
0 3 0 2
1
2 𝑥2 1 2
𝑓𝑌 (𝑦) = [ + 𝑥𝑦 + 𝑥] = (𝑦 + 1), 0≤𝑦≤1
3 2 2 0 3

1
2 1 1
𝑓𝑍 (𝑧) = ∫ 𝑓𝑌𝑍 (𝑦, 𝑧)𝑑𝑦 = ∫ ( + 𝑦 + 𝑧) 𝑑𝑦
0 3 0 2
1
2 1 𝑦2 2
𝑓𝑍 (𝑧) = [ 𝑦 + + 𝑧𝑦] = (𝑧 + 1), 0≤𝑧≤1
3 2 2 0
3

Page 2 of 8
Problem No. 6.22

(a) 𝑈 = 𝑋1, 𝑉 = 𝑋1 + 𝑋2, 𝑊 = 𝑋1 + 𝑋2 + 𝑋3

𝑈 1 0 0 𝑋1
𝒁 = 𝐴𝑿 → [ 𝑉 ] = [1 1 0] [𝑋2 ]
𝑊 1 1 1 𝑋3
➔ |𝐴| = 1

𝑋1 = 𝑈, 𝑋2 = 𝑉 − 𝑋1 = 𝑉 − 𝑈, 𝑋3 = 𝑊 − 𝑉
𝑋1 1 0 0 𝑈
−1 ̃
𝑋 = 𝐴 𝑍 = 𝐴𝑍 → [𝑋2 ] = [−1 1 0] [ 𝑉 ]
𝑋3 0 −1 1 𝑊

𝑓𝑿 (𝑥1 , 𝑥2 , 𝑥3 )
𝑓𝑈,𝑉,𝑊 (𝑢, 𝑣, 𝑤) = |
|𝐴| 𝑿=𝐴̃𝒁

𝑓𝑈𝑉𝑊 (𝑢, 𝑣, 𝑤) = 𝑓𝑿 (𝑥1 , 𝑥2 , 𝑥3 )| = 𝑓𝑿 (𝑢, 𝑣 − 𝑢, 𝑤 − 𝑣)

𝑿=𝐴̃𝒁

(b) Since 𝑋1, 𝑋2 and 𝑋3 are zero-mean , unit-variance Gaussian random variables,
𝑋𝑖 ~𝑁(0,1)
𝑓𝑿 (𝑥1 , 𝑥2 , 𝑥3 ) = 𝑓𝑋1 (𝑥1 )𝑓𝑋2 (𝑥2 )𝑓𝑋3 (𝑥3 )
1 𝑥12 𝑥22 𝑥32
= 3 𝑒− 2 𝑒− 2 𝑒− 2
(√2𝜋)
Therefore,

𝑓𝑈𝑉𝑊 (𝑢, 𝑣, 𝑤) = 𝑓𝑿 (𝑥1 , 𝑥2 , 𝑥3 )|

𝑿=𝐴̃𝒁

1 𝑢2 (𝑣−𝑢)2 (𝑤−𝑣)2
− − −
= 3 𝑒 2 𝑒 2 𝑒 2
(√2𝜋)
(c)
∞ ∞ 𝑢2 (𝑣−𝑢)2 (𝑤−𝑣)2
1 − − −
𝑓𝑉𝑊 (𝑣, 𝑤) = ∫ 𝑓𝑈𝑉𝑊 (𝑢, 𝑣, 𝑤) 𝑑𝑢 = ∫ 3 𝑒 2 𝑒 2 𝑒 2 𝑑𝑢
−∞ −∞ (√2𝜋)

1 − (𝑤−𝑣)2 ∞ 1 −
𝑢2
−
(𝑣−𝑢)2
= 𝑒 2 ∫ 𝑒 2 𝑒 2 𝑑𝑢
2𝜋 −∞ √2𝜋

Page 3 of 8
 1 − (𝑤−𝑣)2 ∞ 1 −𝑣 2 −2𝑢2 +2𝑢𝑣
𝑓𝑉𝑊 (𝑣, 𝑤) = 𝑒 2 ∫ 𝑒 2 𝑑𝑢
2𝜋 −∞ √2𝜋
1 2 1 2
1 − (𝑤−𝑣)2 ∞ 1 − 𝑣
2
−2(𝑢− 𝑣)
2
𝑓𝑉𝑊 (𝑣, 𝑤) = 𝑒 2 ∫ 𝑒 2 𝑑𝑢
2𝜋 −∞ √2𝜋
1 2
1 − (𝑤−𝑣)2 ∞ 1 𝑣2 (𝑢− 𝑣)
2
(𝑣,
𝑓𝑉𝑊 𝑤) = 𝑒 2 ∫ 𝑒 4 𝑒−
−
1 𝑑𝑢
2𝜋 −∞ √2𝜋
1 2
(𝑢− 𝑣)
1 − 2
1 2
√2 1 − (𝑤−𝑣)2 − 𝑣2 ∞ 1 2( )
= × 𝑒 2 𝑒 4∫ 𝑒 √2 𝑑𝑢
1 2π 1
−∞ √2𝜋 ( )
√2
1 1
Since the integrand is the pdf of a Gaussian random variable with mean 𝑣 and variance 2, the
2

integral becomes unity. Therefore,

1 (𝑤−𝑣)2 𝑣 2
− −
𝑓𝑉𝑊 (𝑣, 𝑤) = 𝑒 2 4
2√2𝜋

∞ ∞ (𝑤−𝑣)2 𝑣 2
1
𝑓𝑉 (𝑣) = ∫ 𝑓𝑉𝑊 (𝑣, 𝑤)𝑑𝑤 = ∫ 𝑒− 2
−
4 𝑑𝑤
−∞ −∞ 2√2𝜋
𝑣2 ∞ (𝑤−𝑣)2 𝑣2
1 − 1 − 1 −
= 𝑒 4 ∫ 𝑒 2 𝑑𝑤 = 𝑒 4
2 √𝜋 ⏟−∞ √2𝜋 2√𝜋
=1
(Gaussian with mean 𝑣 and variance 1)

∞ ∞ (𝑤−𝑣)2 𝑣 2
1 − −
𝑓𝑊 (𝑤) = ∫ 𝑓𝑉𝑊 (𝑣, 𝑤)𝑑𝑣 = ∫ 𝑒 2 4 𝑑𝑣
−∞ −∞ 2√2𝜋
2 4
(3𝑣2 +2𝑤2 −4𝑣𝑤) 3(𝑣2 + 𝑤2 − 𝑣𝑤)
∞ 1 ∞ 1 3 3
−
= ∫−∞ 2√2 𝜋 𝑒 4 𝑑𝑣=∫−∞ 2√2 𝜋 𝑒 − 4 𝑑𝑣

2 2 2 2 2
3((𝑣− 𝑤) + 𝑤2 ) 2 2 2 (𝑣− 𝑤) 𝑤2
3 9 (𝑣− 𝑤) + 𝑤2 − ( 3 )−( )
∞ ∞ 3 9 ∞ 4
1 1 − 1 6
= ∫−∞ 2√2 𝜋 𝑒 − 4 𝑑𝑣 = ∫−∞ 2√2 𝜋 𝑒 4/3 𝑑𝑣 = ∫−∞ 2√2 𝜋 𝑒 3 𝑑𝑣

Page 4 of 8
 2 2
(𝑣− 𝑤)
− 3
2
𝑤2 ∞ 2
1 1 2(√ )
= 𝑒− 6 ∫ 𝑒 3
𝑑𝑣
√2𝜋 × 3 ⏟−∞ √2𝜋 × 2/3
=1
2 2
(Gaussian with mean 𝑤 and variance )
3 3
2 2
Since integrand is the pdf of Gaussian random variable with mean 3 𝑤 and variance 3, the

integral becomes unity. Therefore,

1 𝑤2
−
𝑓𝑊 (𝑤) = 𝑒 (2)(3) , −∞<𝑤 <∞
√2𝜋 × 3
Thus, 𝑊 is a Gaussian random variable with mean zero and variance of 3.

Page 5 of 8
Problem No. 7.1

𝑊 = 𝑋 + 𝑌 + 𝑍, and 𝑋, 𝑌 and 𝑍 are zero-mean, unit-variance random variables.

1 1 1
𝐶𝑂𝑉(𝑋, 𝑌) = 2, 𝐶𝑂𝑉(𝑌, 𝑍) = − 4, 𝐶𝑂𝑉(𝑋, 𝑍) = 2

(a)
𝐸[𝑊] = 𝐸[𝑋 + 𝑌 + 𝑍] = 𝐸[𝑋] + 𝐸[𝑌] + 𝐸[𝑍] = 0
𝑉𝐴𝑅[𝑊] = 𝑉𝐴𝑅[𝑋 + 𝑌 + 𝑍]
= 𝑉𝐴𝑅[𝑋] + 𝑉𝐴𝑅[𝑌] + 𝑉𝐴𝑅[𝑍] + (2)𝐶𝑂𝑉(𝑋, 𝑌) + (2)𝐶𝑂𝑉(𝑋, 𝑍)
+ (2)𝐶𝑂𝑉(𝑌, 𝑍)
1 1 1
= 1 + 1 + 1 + (2) ( ) + (2) ( ) + (2) (− ) = 4.5
2 2 4
(b) 𝑋, 𝑌 and 𝑍 are uncorrelated.
𝐶𝑂𝑉(𝑋, 𝑌) = 𝐶𝑂𝑉(𝑌, 𝑍) = 𝐶𝑂𝑉(𝑋, 𝑍) = 0
𝐸[𝑊] = 0
𝑉𝐴𝑅[𝑊] = 𝑉𝐴𝑅[𝑋] + 𝑉𝐴𝑅[𝑌] + 𝑉𝐴𝑅[𝑍] = 3

Page 6 of 8
Problem No. 7.7
𝑋 and 𝑌 independent exponential random variables with parameters 𝛼 = 2, and 𝛽 = 10,
respectively, and 𝑍 = 𝑋 + 𝑌
(a)
𝑓𝑋 (𝑥) = 𝛼𝑒 −𝛼𝑥 = 2𝑒 −2𝑥 𝑥 ≥0
𝑓𝑌 (𝑦) = 𝛽𝑒 −𝛽𝑥 = 10𝑒 −10𝑥 𝑥 ≥0
Z=X+Y
Φ𝑍 (𝜔) = 𝐸(𝑒 𝑗𝜔𝑧 ) = 𝐸(𝑒 𝑗𝜔(𝑥+𝑦) ) = 𝐸(𝑒 𝑗𝜔𝑥 )𝐸(𝑒 𝑗𝜔𝑦 )
∞ ∞
𝛼 ∞
𝐸(𝑒 𝑗𝜔𝑥 ) = ∫ 𝛼𝑒 −𝛼𝑥 × 𝑒 𝑗𝜔𝑥 𝑑𝑥 = ∫ 𝛼𝑒 −(𝛼−𝑗𝜔)𝑥 𝑑𝑥 = [−𝑒 −(𝛼−𝑗𝜔)𝑥 ]0
0 0 𝛼 − 𝑗𝜔
2
=
2 − 𝑗𝜔
Similarly,
10
𝐸(𝑒 𝑗𝜔𝑦 ) = 10−𝑗𝜔

Therefore,

𝛼 𝛽 20
Φ𝑍 (𝜔) = ( )( )=
𝛼 − 𝑗𝜔 𝛽 − 𝑗𝜔 (2 − 𝑗𝜔)(10 − 𝑗𝜔)

(b) Using partial fraction

20 𝐴 𝐵
=( )+( )
(2 − 𝑗𝜔)(10 − 𝑗𝜔) 2 − 𝑗𝜔 10 − 𝑗𝜔
𝛽 − 𝛼 −1 5
𝐴=( ) =
𝛼𝛽 2
𝛼 − 𝛽 −1 5
𝐵=( ) =−
𝛼𝛽 2
5 5
−2 5 2 5 10
Φ𝑍 (𝜔) = ( 2 )+ ( )= ( )− ( )
2 − 𝑗𝜔 10 − 𝑗𝜔 4 2 − 𝑗𝜔 20 10 − 𝑗𝜔

5 1
𝑓𝑍 (𝑧) = 𝑒 −2𝑧 − 𝑒 −10 𝑧 , 𝑧 > 0
4 4

Page 7 of 8
Problem No. 7.8
𝑍 = 3𝑋 − 7𝑌, 𝑋 and 𝑌 are independent random variables
(a)
Φ𝑍 (𝜔) = 𝐸[𝑒 𝑗𝜔(3𝑋−7𝑌) ] = 𝐸[𝑒 𝑗𝜔3𝑋 ]𝐸[𝑒 −𝑗𝜔7𝑌 ]
Φ𝑍 (𝜔) = Φ𝑋 (3𝜔)Φ𝑌 (−7𝜔)
(b)
1 𝑑 1 𝑑
𝐸[𝑍] = Φ𝑍 (𝜔)| = [Φ (3𝜔)Φ𝑌 (−7𝜔)]|
𝑗 𝑑𝜔 𝜔=0 𝑗 𝑑𝜔 𝑋 𝜔=0

3 ′ −7
= Φ𝑋 (3𝜔)Φ𝑌 (−7𝜔)| + Φ𝑋 (3𝜔)Φ𝑌′ (−7𝜔)|
𝑗 𝜔=0
𝑗 𝜔=0

𝐸[𝑍] = 3𝐸[𝑋] − 7𝐸[𝑌]

𝐸[𝑍 2 ] = −Φ𝑍′′ = −[(3)2 Φ𝑋′′ (3𝜔)Φ𝑌 (−7𝜔) + 2Φ𝑋′ (3𝜔)(3)(−7)Φ𝑌′ (−7 𝜔)
+ (−7)2 Φ𝑋 (3𝜔)Φ𝑌′′ (−7𝜔)]|𝜔=0

𝐸[𝑍 2 ] = (3)2 𝐸[𝑋 2 ] − 2(3)(−7) × 𝑗𝐸[𝑋] × 𝑗𝐸[𝑌] + (−7)2 𝐸[𝑌 2 ]

= 9𝐸[𝑋 2 ] − 42𝐸[𝑋]𝐸[𝑌] + 49𝐸[𝑌 2 ]

𝑉𝐴𝑅[𝑍] = 𝐸[𝑍 2 ] − (𝐸[𝑍])2

= 9𝐸[𝑋 2 ] − 42𝐸[𝑋]𝐸[𝑌] + 49𝐸[𝑌 2 ] − (9(𝐸[𝑋])2 + 49(𝐸[𝑌])2
− 42 𝐸[𝑋]𝐸[𝑌])
= 9𝐸[𝑋 2 ] - 9(𝐸[𝑋])2 + 49𝐸[𝑌 2 ] − 49(𝐸[𝑌])2
= (9)𝑉𝐴𝑅[𝑋] + (49)𝑉𝐴𝑅[𝑌]

Page 8 of 8