Reference Answers For Assignment4
Reference Answers For Assignment4
4.4 The RV is x is𝑵(𝜼, 𝝈) and 𝑷{𝜼 − 𝒌𝝈 < 𝒙 < 𝜼 + 𝒌𝝈} = 𝒑𝒌. (a) Find𝒑𝒌 for
k=1,2 and 3. (b) Find k for 𝒑𝒌 = 𝟎. 𝟗, 𝟎. 𝟗𝟗 𝒂𝒏𝒅 𝟎. 𝟗𝟗𝟗. (c) If𝑷{𝜼 − 𝒛𝒖𝝈 < 𝒙 < 𝜼
+ 𝒛𝒖𝝈} = 𝜸, express 𝒛𝒖 in terms of 𝜸.
Answer:
(𝜼+𝒌𝝈)−𝜼 (𝜼−𝒌𝝈)−𝜼
(a) 𝑝𝑘 = 𝐺 ( )−𝐺( ) = 𝐺(𝑘) − 𝐺(−𝑘) = 𝐺(𝑘) − {1 − 𝐺(𝑘)}
𝝈 𝝈
1
𝑝𝑘 = 2𝐺(𝑘) − 1 = 2 {𝐺(𝑘) − } = 2𝑒𝑟𝑓(𝑘)
2
From table 3-1 (Papoulis, 1991)
For 0.45 erf(k) or 𝑈 is lie between value range Ua=0.44520 (for 𝑥𝑎 =1.60) and
0.45053 (for 𝑥𝑏 =1.65).
1.65 − 1.60
𝑥𝑢 ≅ 1.6 + (0.45 − 0.44520) = 1.645
0.45053 − 0.44520
For 0.495 erf(k) or 𝑈 is lie between value range Ua=0.49461 (for 𝑥𝑎 =2.55) and
0.49534 (for 𝑥𝑏 =2.6).
2.6 − 2.55
𝑥𝑢 ≅ 2.55 + (0.495 − 0.49461) = 2.5767
0.49534 − 0.49461
For 0.4995 erf(k) or 𝑈 is lie between value range Ua=0.49865 (for 𝑥𝑎 =3) and
0.5 (for 𝑥𝑏 =4).
4−3
𝑥𝑢 ≅ 3 + (0.49534 − 0.49865) = 3.6296
0.5 − 0.49865
This last interpolation take larger distance between 3 to 4 which is equal to one
and create inaccurate calculation because the linear curve did not same with
exponential curve (it is only approximate the and predict the value by linear
interpolation), since we do not have value above 3 which available in the table.
Assumption for erf(4) is 0.5.
If we use the matlab to calculate the integral function for error function
Then it will result exact value as on the table or the value calculate by integrate
the function of exponential (exponential curve not linear).
For the correct answer for pk=0.999 or erf(k)= 0.4995 is 3.3 ( 3.26-3.33)
(𝜂+𝑧 𝜎)−𝜂
𝑢 (𝜂−𝑧 𝜎)−𝜂
𝑢
(c) 𝑃{𝜂 − 𝑧𝑢 𝜎 < 𝑥 < 𝜂 + 𝑧𝑢 𝜎} = 𝐺 ( )−𝐺( )=
𝜎 𝜎
𝐺(𝑧𝑢 ) − 𝐺(−𝑧𝑢 ) = 𝐺(𝑧𝑢 ) − {1 − 𝐺(𝑧𝑢 )} = 2𝐺(𝑧𝑢 ) − 1 = 𝛾
𝑃{𝜂 − 𝑧𝑢 𝜎 < 𝑥 < 𝜂 + 𝑧𝑢 𝜎} = 2𝐺(𝑧𝑢 ) − 1 = 𝛾
𝐺(𝑧𝑢 ) = (1 + 𝛾)/2
𝑢 = (1 + 𝛾)/2
4.5 Find 𝒙𝒖 for u=0.1, 0.2, … 0,9 (a) if x is uniform in the interval (0,1); (b)
𝒊𝒇 𝒇(𝒙) = 𝒆−𝟐𝒙 𝑼(𝒙).
Answer:
(a) 𝐹(𝑥) = 𝑥 𝑓𝑜𝑟 0 ≤ 𝑥 ≤ 1; hence 𝑢 = 𝐹(𝑥𝑢 ) = 𝑥𝑢
(b) 𝐹(𝑥) = 1 − 𝑒 −2𝑥 𝑓𝑜𝑟 𝑥 ≥ 0; hence 𝑢 = 1 − 𝑒 −2𝑥𝑢 from this we can find 𝑥𝑢
By adding 𝑒 −2𝑥𝑢 on both sides then we have 𝑒 −2𝑥𝑢 + 𝑢 = 1 − 𝑒 −2𝑥𝑢 + 𝑒 −2𝑥𝑢
𝑒 −2𝑥𝑢 + 𝑢 = 1
Then minus u for both sides we have 𝑒 −2𝑥𝑢 + 𝑢 − 𝑢 = 1 − 𝑢
By adding log natural in both sides
𝑙𝑛𝑒 −2𝑥𝑢 = ln (1 − 𝑢)
−2𝑥𝑢 = ln (1 − 𝑢)
1
𝑥𝑢 = − ln(1 − 𝑢)
2
1
𝑥0.1 = − ln(0.9) = −0.5 ∗ (−0.1054) = 0.0527;
2
1
𝑥0.2 = − ln(0.8) = −0.5 ∗ (−0.2231) = 0.1116;
2
1
𝑥0.3 = − ln(0.7) = −0.5 ∗ (−0.3567) = 0.1783
2
𝑥0.4 = 0.2554;𝑥0.5 = 0.3466; 𝑥0.6 = 0.4581
𝑥0.7 = 0.6020; 𝑥0.8 = 0.847;𝑥0.9 = 1.1513
4.6 We measure for resistance R of each resistor in a production line and we accept
only the units the resistance of which is between 96 and 104 ohms. Find the
percentage of the accepted units (a) if R is uniform between 95 and 105 ohms; (b)
if R is normal with 𝜼 = 𝟏𝟎𝟎 𝒂𝒏𝒅 𝝈 = 𝟐 𝒐𝒉𝒎𝒔.
Answer:
Percentage of units between 96 and 104 ohms equals 100p where 𝑝 = 𝑃{96 < 𝑅 <
104} = 𝐹(104) − 𝐹(96)
1 1 1
(a) Because R uniform between 95 and 105 ohms then 𝑥 −𝑥 = 105−95 = 10 = 0.1
2 1
𝐹(𝑅) = 0.1(𝑅 − 95)𝑓𝑜𝑟 95 ≤ 𝑅 ≤ 105. Hence, 𝑝 = 0.1(104 − 95) −
0.1(96 − 95) = 0.8
(b) Random variable called as normal or gaussian if the density creates normal
shape, movable and scalable. This curve creates bell shape which is symmetric
to line 𝒙 = 𝜼
𝑥−𝜂 𝑥−𝜂
𝑝 = 𝑃{96 < 𝑅 < 104} = 𝐹(𝑥) = 𝐺 ( )−𝐺( )
𝜎 𝜎
104 − 100 96 − 100
=𝐺( )−𝐺( ) = 𝐺(2) − 𝐺(−2)
2 2
1
= 𝐺(2) − {1 − 𝐺(2)} = 2𝐺(2) − 1 = 2 ∗ {𝐺(2) − }
2
= 2 erf(2) = 2 ∗ 0.47726 = 0.9545
Bibliography
Papoulis, A. (1991). Probability, Random Variables and Stochastic Processes (3rd Edition ed.). St. Louis,
New York, United State: McGraw Hill, Inc.