National Central University

Engineering Math (ME2001-B) Jan 8, 2023

Final Exam (10:00 – 12:50)

 Close books and Allow 2 page of Notes
 Do NOT use calculators, neither do any types of electronic devices (不可使用計算器與任何電子元件如手機、
Pad 等)

𝑑𝑦
Problem 1 (10%) An ODE is given by: + 𝑦 = 𝑡 with initial condition: 𝑦(0) = 1. Find 𝑦 using the
𝑑𝑡

method of undetermined coefficients.

𝑑𝑦
Sol. 𝑦 = 𝑦ℎ + 𝑦𝑝 ; 𝑦ℎ satisfies + 𝑦 = 0.
𝑑𝑡

Let 𝑦ℎ = 𝑒 𝜆𝑡 , then 𝜆 + 1 = 0, 𝜆 = −1 ∴ 𝑦ℎ = 𝑐𝑒 −𝑡 (4%)

Let 𝑦𝑝 = 𝑎𝑡 + 𝑏. Substitute it into the ODE, then 𝑎 + 𝑎𝑡 + 𝑏 = 𝑡
∴ 𝑎 = 1, 𝑏 = −1, 𝑦𝑝 = 𝑡 − 1.
Combine 𝑦ℎ with 𝑦𝑝 , we have 𝑦 = 𝑐𝑒 −𝑡 + 𝑡 − 1. By 𝑦(0) = 1, we have 𝑦 = 2𝑒 −𝑡 + 𝑡 − 1. QED

Problem 2 (25%)
A mechanical oscillation (vibration) system is described by the following ODE:
𝑦 ′′ + 𝑘𝑦 = cos(𝜔𝑡), 𝑘 > 0.
The solution can be expressed as 𝑦(𝑡) = 𝑦ℎ (𝑡) + 𝑦𝑝 (𝑡), where 𝑦ℎ and 𝑦𝑝 are the homogeneous and
nonhomogeneous solutions respectively. The nonhomogeneous solution 𝑦𝑝 (𝑡) is plotted on the figure
below.

(a) (4%) Briefly explain the physical meanings of 𝑦𝑝 (𝑡) (or the system).
(b) (4%) Find the nonhomogeneous solution 𝑦𝑝 (𝑡).
(c) (4%) Find the value 𝜔.
(d) (4%) Find the value of 𝑘.
(e) (4%) Find the natural frequency of the system 𝑓0 = 𝜔0 /(2𝜋).
(f) (5%) Suppose the ICs are given with 𝑦(0) = 𝑦 ′ (0) = 0. Find the homogeneous solution 𝑦(𝑡).
Sol.
(a) Based on the figure, the amplitude increases with time (is proportional to time), and it approaches
infinity as time 𝑡 → ∞. It means resonance occurs.
(b) Since resonance occurs, then, 𝜔0 = 𝜔 = 2𝜋, the nonhomogeneous solution is modified by a factor
𝑡. That is, 𝑦𝑝 is modified as follows.
𝑦𝑝 = 𝑡[𝑎𝑐𝑜𝑠(𝜔𝑡) + 𝑏𝑠𝑖𝑛(𝜔𝑡)]. Substitute into Eq. (1), we have 𝑎 = 0, 𝑏 = 1. Then
𝑦𝑝 = 𝑡 ∙ 𝑠𝑖𝑛(𝜔𝑡). (1)
2𝜋
(c) Following the figure, the period of 𝑠𝑖𝑛(𝜔𝑡) is 𝑇 = 1 → 𝜔 = = 2𝜋.
𝑇

(d) Following (b), we have 𝜔02 = 4𝜋 2 = 𝑘/1, i.e., 𝑘 = 4𝜋 2 .

𝜔0
(e) Follow (f), it yields 𝑓0 = = 1.
2𝜋

(f) Let 𝑦ℎ = 𝑒 𝜆𝑡 , and we obtain the homogeneous solution 𝑦ℎ is as follows .

𝑦ℎ = 𝐴𝑐𝑜𝑠(𝜔0 𝑡) + 𝐵𝑠𝑖𝑛(𝜔0 𝑡), 𝜔02 = 𝑘 (2)

Combine (1) with (2),

𝑦 = 𝑦ℎ + 𝑦𝑝 = 𝐴𝑐𝑜𝑠(𝜔0 𝑡) + 𝐵𝑠𝑖𝑛(𝜔0 𝑡) + 𝑡 ∙ 𝑠𝑖𝑛(𝜔𝑡)
Using the ICs 𝑦(0) = 𝑦 ′ (0) = 0 and apply Eq. (3), we have 𝐴 = 𝐵 = 0. Consequently,
𝑦(𝑡) = 𝑡 ∙ 𝑠𝑖𝑛(𝜔𝑡) QED

Problem 3 (10%)
(a) (5%) An ODE is written as: 𝐿(𝑦) ≜ 𝑦 ′′′ + 23𝑦 ′′ + 𝑦 = 𝑟1 (𝑥). Let 𝑦1 (𝑥) be the solution of 𝐿(𝑦) =
𝑟1 (𝑥) and 𝑦2 (𝑥) be the solution of 𝐿(𝑦) = 𝑟2 (𝑥) . Find the solution of 𝐿(𝑦) = 4𝑟1 (𝑥) + 3𝑟2 (𝑥) in
terms of 𝑦1 and 𝑦2 .
(b) (5%) Does the following 3 functions: 𝑓(𝑡), 𝑔(𝑡), and 𝑧(𝑡) = 2𝑓(𝑡) + 3𝑔(𝑡) form a linearly
independent basis or not.
Sol.
(a) Use the linearity (superposition), we have
𝐿(4𝑦1 + 3𝑦2 ) = 4𝐿(𝑦1 ) + 3𝐿(𝑦2 ) = 4𝑟1 (𝑥) + 3𝑟2 (𝑥)
y=4𝑦1 + 3𝑦2 QED
(b) Since 𝑧(𝑡) is a linear combination of 𝑓(𝑡) and 𝑔(𝑡). Consequently, they are linearly dependent
(i.e. Not linearly independent). Alternatively, you can apply the Wronskian of the 3 functions to draw the
same conclusion.

Problem 4 (10%)
−3 1
Let 𝑦⃑ = [𝑦1 , 𝑦2 ]𝑇 . A system of ODEs is written as: 𝑦⃑ ′ = 𝐴𝑦⃑ with 𝐴 = [ ]. Let 𝜆1 and 𝜆2 are
1 −3
1 1
the eigenvalues of 𝐴 with |𝜆1 | ≤ |𝜆2 |, In addition, let 𝑥⃑1 = ( ) and 𝑥⃑2 = ( ) are the eigenvectors
𝑥21 𝑥22
correspond to 𝜆1 and 𝜆2 , respectively.
(a) (2%) Find the values of 𝜆1 and 𝜆2 .
(b) (2%)Find 𝑥21 and 𝑥22 .
(c) (4%) Suppose the initial conditions are given by 𝑦1 (0) = 0, 𝑦2 (0) = 2, find the solution 𝑦⃑.
(d) (2%) Is the system stable, unstable, or neutral stable?
Sol.
−3 1
(a) 𝐴 = [ ], 𝜆1 = −2, 𝜆2 = −4,
1 −3
1 1
(b) 𝑥⃑1 = ( ), 𝑥⃑2 = ( )
1 −1
1 1
(c) Then 𝑦⃑ = 𝑐1 𝑒 −2𝑡 ( ) + 𝑐2 𝑒 −4𝑡 ( ), Apply ICs 𝑦1 (0) = 0, 𝑦2 (0) = 2, we obtain
1 −1
𝑦1 0 𝑐 +𝑐 𝑦1 −2𝑡 −4𝑡
(𝑦 ) = ( ) = (𝑐1 − 𝑐2 ) then 𝑐1 = 1, 𝑐2 = −1 ∴ (𝑦 ) = (𝑒 −2𝑡 − 𝑒 −4𝑡 )
2 2 1 2 2 𝑒 +𝑒
0
(d) Since lim 𝑦⃑ → ( ), the system is unstable. QED
𝑡→∞ 0

Problem 5 (10%) An ODE is given by: 𝑦 ′′ + 𝑦 ′ + 2𝑥𝑦 = 0.

(a) (4%) Point 𝑥 = 0 is a regular point , regular singular point, or irregular singular point?
(b) (6%) Find the first six terms in the power series solution of the ODE
Sol.
(a) Regular Point,
(b) Follow (a), we can apply power series 𝑦 = ∑∞ 𝑛=0 𝑎𝑛 𝑥
𝑛
(Or Frobenius series 𝑦 = ∑∞ 𝑛=0 𝑎𝑛 𝑥
𝑛+𝑟

to solve the ODE.

We have
𝑦 = ∑∞ 𝑛=2[(𝑛 + 1)(𝑛)𝑎𝑛+1 + 𝑛𝑎𝑛 + 2𝑎𝑛−2 ]𝑥
𝑛−1
+ 2𝑎2 + 𝑎1 = 0,
Then
1 −1 2
∴ 𝑎2 = (2) 𝑎1 , 𝑎𝑛+1 = (𝑛+1) 𝑎𝑛 − 𝑛(𝑛+1) 𝑎𝑛−2

1 1
𝑛 = 2, 𝑎3 = (6) 𝑎1 − (3) 𝑎0

−5 1
𝑛 = 3, 𝑎4 = ( 24 ) 𝑎1 + (12) 𝑎0

11 1
𝑛 = 4, 𝑎5 = (120) 𝑎1 − (60) 𝑎0

1 1 1 −5 1 11 1
𝑦 = 𝑎0 + 𝑎1 𝑥 − (2) 𝑎1 𝑥 2 + [(6) 𝑎1 − (3) 𝑎0 ] 𝑥 3 + [( 24 ) 𝑎1 + (12) 𝑎0 ] 𝑥 4 + [(120) 𝑎1 − (60) 𝑎0 ] 𝑥 5 +

⋯ QED

Problem 6 (10%)
Denote 𝛿(𝑡) as the Dirac’s Delta function. An ODE is written as 𝑦 ′′ + 4𝑦 ′ + 3𝑦 = 𝛿(𝑡 − 2) with initial
conditions 𝑦(0) = 𝑦 ′ (0) = 0. Solve the ODE by Laplace transform:
Apply the formula:
𝐿(𝑦(𝑡)) = 𝑌(𝑠)
𝐿(𝑦 ′ (𝑡)) = 𝑠𝑌(𝑠) − 𝐾0
𝐿(𝑦 ′′ (𝑡)) = 𝑠 2 𝑌(𝑠) − 𝑠𝐾0 − 𝐾1
𝐿(𝑟(𝑡)) = 𝑅(𝑠)
Substitute them into the ODE: 𝑦 ′′ + 𝑎𝑦 ′ + 𝑏𝑦 = 𝑟(𝑡), then we have
1
𝑌(𝑠) = [(𝑠 + 𝑎)𝐾0 + 𝐾1 ]𝑄(𝑠) + 𝑅(𝑠)𝑄(𝑠) = 2 𝑅(𝑠)
(𝑠 + 𝑎𝑠 + 𝑏)
Then
𝑒 −2𝑠 1/2 −1/2
𝑌(𝑠) = (𝑠+3)(𝑠+1) = 𝑒 −2𝑠 [𝑠+1 + ] Apply t-shift theorem
𝑠+3

0, 0 ≤ 𝑡 < 2
𝑦(𝑡) = 𝐿−1 (𝑌(𝑠)) = { −(𝑡−2) QED
0.5 ∙ (𝑒 − 𝑒 −3(𝑡−2) ), 𝑡 ≥ 2

Problem 7 (10%)
Let 𝑓(𝑡) be a periodic function, 𝑓(𝑡) = 𝑓(𝑡 + 𝑝) with period 𝑝 . Denote L[𝑓(𝑡)] as the Laplace
𝑝
∫0 𝑒 −𝑠𝑡 𝑓(𝑡)𝑑𝑡
transform of 𝑓(𝑡). Prove 𝐿[𝑓(𝑡)] = .
1−𝑒 −𝑠𝑝

Sol.
∞ 𝑡=𝑝 𝑡=2𝑝
Proof. 𝐿[𝑓(𝑡)] = ∫0 𝑒 −𝑠𝑡 𝑓(𝑡)𝑑𝑡 = ∫𝑡=0 𝑒 −𝑠𝑡 𝑓(𝑡)𝑑𝑡 + ∫𝑡=𝑝 𝑒 −𝑠𝑡 𝑓(𝑡)𝑑𝑡 + ⋯
𝑡=𝑝
Denote ∫𝑡=0 𝑒 −𝑠𝑡 𝑓(𝑡)𝑑𝑡 = 𝑓1 (𝑡)
Let 𝑢 = 𝑡 − 𝑝 in the second integral, then
𝑡=2𝑝 𝑢=𝑝 𝑢=𝑝
∫𝑡=𝑝 𝑒 −𝑠𝑡 𝑓(𝑡)𝑑𝑡 = ∫𝑡=0 𝑒 −𝑠(𝑢+𝑝) 𝑓(𝑢)𝑑𝑢 = 𝑒 −𝑠𝑝 ∫𝑡=0 𝑒 −𝑠𝑢 𝑓(𝑢)𝑑𝑢 = 𝑒 −𝑠𝑝 𝑓1 (𝑡)

𝑝
∫0 𝑒 −𝑠𝑡 𝑓(𝑡)𝑑𝑡
Therefore, 𝐿[𝑓(𝑡)] = 𝑓1 (𝑡) + 𝑒 −𝑠𝑝 𝑓1 (𝑡) + 𝑒 −2𝑠𝑝 𝑓1 (𝑡) + ⋯ = 1−𝑒 −𝑠𝑝
. QED

Problem 8 (5%)
The functions 𝑦1 (𝑥) and 𝑦2 (𝑥) are given by 𝑦1 (𝑥) = 3𝑥 2 − 1 and 𝑦2 (𝑥) = 𝑥, respectively.
(a) (3%) Is 𝑦1 (𝑥) orthogonal to 𝑦2 (𝑥) on interval 𝐼 = [−1, 1]? (please gives your derivation)
(a) (2%) Is 𝑦1 (𝑥) orthogonal to 𝑦2 (𝑥) on interval 𝐼 = [0, 1]? (please gives your derivation)
Sol. Let 𝑟(𝑥) = 1, then
1 1
(a) ∫−1 r(x) ∙ y1 y2 dx = ∫−1(3𝑥 3 − 𝑥) 𝑑𝑥 = 0. Consequently 𝑦1 is orthogonal to 𝑦2 .

1
(b) ∫0 𝑟(𝑥) ∙ 𝑦1 𝑦2 𝑑𝑥 = −1/4 ≠ 0. Consequently 𝑦1 is NOT orthogonal to 𝑦2 .

Problem 9 (10%) Expand 𝑠𝑖𝑛5 (𝑡) by Fourier series. (hint: using Euler formula)
Sol. (a) Based on Euler formula: 𝑒 ±𝑖𝑛𝜃 = cos(𝑛𝜃) ± 𝑖sin(𝑛𝜃), it follows that
1 5
𝑠𝑖𝑛5 (𝑡) = [2𝑖 (𝑒 𝑖𝑡 − 𝑒 −𝑖𝑡 ) ] (1).
Using binomial expansion: (𝑎 + 𝑏)5 = 𝑎5 + 5𝑎4 𝑏 + 10𝑎3 𝑏 2 + 10𝑎2 𝑏 3 + 5𝑎𝑏 4 + 𝑏 5 .
Substitute the above formula into Eq. (1), then
1
𝑠𝑖𝑛5 (𝑡) = 32𝑖 (𝑒 5𝑖𝑡 − 5𝑒 3𝑖𝑡 + 10𝑒 𝑖𝑡 − 10𝑒 −𝑖𝑡 + 5𝑒 −3𝑖𝑡 − 𝑒 −5𝑖𝑡 )

5 5 1
= 8 sin(𝑡) − 16 sin(3𝑡) + 16 sin(5𝑡) QED

Problem 10 (10%)
A function is given by: 𝑓(𝑥) = 𝑠𝑖𝑛(𝑥) , 𝑥 ∈ [0, 𝜋]. A function 𝑔(𝑥) is generated by extending 𝑓(𝑥)
from 𝑥 ∈ [0, 𝜋] to 𝑥 ∈ (−∞, ∞) as shown in the figure below.

Let 𝑇 be the period of 𝑔(𝑥). Also define radian frequency 𝜔𝑛 = 𝑛𝜔0 , 𝑛 = 1, 2, ⋯ , 𝜔0 = 2𝜋/𝑇. Find
the (half-range) Fourier series of 𝑔(𝑥). That is
𝑔(𝑥) = 𝑎0 + ∑∞ 𝑛=1[𝑎𝑛 𝑐𝑜𝑠(𝜔𝑛 𝑥) + 𝑏𝑛 𝑠𝑖𝑛(𝜔𝑛 𝑥)] , where 𝜔𝑛 = 𝑛𝜔0 , 𝑛 = 1, 2, ⋯ , 𝜔0 = 2𝜋/𝑇, and
1 𝑇/2
𝑎0 = 𝑇 ∫−𝑇/2 𝑓(𝑥) 𝑑𝑥,

2 𝑇/2
𝑎𝑛 = 𝑇 ∫−𝑇/2 𝑓(𝑥) 𝑐𝑜𝑠(𝜔𝑛 𝑥)𝑑𝑥, 𝑛 = 1, 2, ⋯,

2 𝑇/2
𝑏𝑛 = 𝑇 ∫−𝑇/2 𝑓(𝑥) 𝑠𝑖𝑛((𝜔𝑛 𝑥))𝑑𝑥; 𝑛 = 1,2, ⋯

(a) (2%) What is the value of 𝑇

(b) (4%) Find the coefficients 𝑏7 and 𝑏9 .
(4) (4%)Find the coefficients 𝑎0 and 𝑎1 .

Sol. (a) From the figure, 𝑇 = 2𝜋

(b) Since 𝑔(𝑥) is an even function, it follows that the coefficients of the sine are all zeros, 𝑏𝑘 = 0 for
any 𝑘.
1 𝜋
(c) 𝑎0 = 𝜋 ∫0 sin(𝑥) 𝑑𝑥 = 2/𝜋;

0, 𝑛=1
2 𝜋 𝑛]
𝑎𝑛 = ∫ sin(𝑥) cos(𝑛𝑥) 𝑑𝑥 = {−2[1 + (−1)
𝜋 0 ,𝑛 ≠ 1
𝜋(𝑛2 − 1)
To sum up, 𝑎0 = 2/𝜋, 𝑎1 = 0 QED