WTW 238

– SUMMARY –
Theme 1: Linear systems of differential equations

1.1 The eigenvalue eigenvector method

To solve the 𝑋̅ ′ = 𝐴𝑋̅ homogenous system:

• 𝐴 is a 𝑛 × 𝑛 matrix
• The solution is in the form 𝑋̅ = 𝐾
̅ 𝑒 𝜆𝑡

Definitions
 𝜆 is an eigenvalue of 𝐴 if |𝑨 − 𝝀𝑰| = 𝟎
̅ so that (𝑨 − 𝝀𝑰)𝑲
 An eigenvector associated with 𝜆 is a non-zero vector 𝐾 ̅ =𝟎
̅

Method

Step 1: Determine the eigenvalues (𝜆1 , 𝜆2 , 𝜆3 , … , 𝜆4 ) of 𝐴 using |𝑨 − 𝝀𝑰| = 𝟎

Step 2: ̅1 , 𝐾
Try to find the linearly independent eigenvectors (𝐾 ̅2 , 𝐾
̅3 , … , 𝐾
̅𝑛 ) associated
̅ =𝟎
with each of the eigenvalues using (𝑨 − 𝝀𝑰)𝑲 ̅

Step 3: If linearly independent eigenvectors are found, the linearly independent

̅𝟏 = 𝑲
solutions are 𝑿 ̅ 𝟏 𝒆𝝀𝟏 𝒕, 𝑿
̅𝟐 = 𝑲
̅ 𝟐 𝒆𝝀𝟐 𝒕 , …., 𝑿
̅𝒏 = 𝑲
̅ 𝒏 𝒆𝝀𝒏 𝒕 [fundamental set

of solutions]
Step 4: The general solution of 𝑋̅ ′ = 𝐴𝑋̅ is a linear combination of the fundamental set
of solutions:
 𝑋̅(𝑡) = 𝑐1 𝑋̅1 + 𝑐2 𝑋̅2 + 𝑐3 𝑋̅3 + ⋯ + 𝑐𝑛 𝑋̅𝑛
̅ (𝒕) = 𝒄𝟏 𝑲
 𝑿 ̅ 𝟏 𝒆𝝀𝟏 𝒕 + 𝒄𝟐 𝑲
̅ 𝟐 𝒆𝝀𝟐 𝒕 + 𝒄𝟑 𝑲
̅ 𝟑 𝒆𝝀𝟑 𝒕 + ⋯ + 𝒄𝒏 𝑲
̅ 𝒏 𝒆𝝀𝒏 𝒕
Step 5: To solve an initial value problem (IVP), substitute initial values to find
𝒄𝟏 , 𝒄𝟐 , 𝒄𝟑 , … , 𝒄𝒏 to find the particular solution.

Some important notes:

To determine whether a solution forms a fundamental set or not:
• Wronskian → 𝑊 (𝑋̅1 , 𝑋̅2 , … , 𝑋̅𝑛 ) ≠ 0
• Definition → 𝑋̅(𝑡) = 0; 𝑐1 = 𝑐2 = 𝑐3 = ⋯ = 𝑐𝑛 = 0
• Distinct eigenvalues, each with its own eigenvector
Definition: linear independence
̅𝟏 , 𝒗
Vectors 𝒗 ̅𝟐 , 𝒗
̅𝟑 , … , 𝒗
̅𝒏 are linearly independent if 𝒄𝟏 = 𝒄𝟐 = 𝒄𝟑 = ⋯ = 𝒄𝒏 = 𝟎 is the only
̅𝟏 + 𝒄𝟐 𝒗
solution of the system, 𝒄𝟏 𝒗 ̅𝟐 + 𝒄𝟑 𝒗
̅𝟑 + ⋯ + 𝒄𝒏 𝒗
̅𝒏

∴ Two vectors are linear independent if neither one is a multiple of the other.

̅ is an eigenvector associated with the eigenvalue 𝜆, 𝑐𝐾

∴ If 𝐾 ̅ is also an eigenvector

associated with 𝜆 for any constant 𝑐 ≠ 0

A few more notes:

 If |𝐴| ≠ 0, 𝐴−1 exists.
̅ , 1 is an eigenvalue of 𝐴−1 with
 If 𝜆 is an eigenvalue of 𝐴 with the associated eigenvector 𝐾 𝜆
̅.
the associated eigenvector 𝐾
̅ , 𝜆2 is an eigenvalue of 𝐴2 with
 If 𝜆 is an eigenvalue of 𝐴 with the associated eigenvector 𝐾
̅.
the associated eigenvector 𝐾
̅ = 𝜆𝐾
 For proofs: starts with 𝐴𝐾 ̅

 To determine eigenvalues, look out for the following:

𝒂 𝟎 𝟎 𝒂 𝑒 𝑓
𝒂 𝑐
𝐴=[ 𝑒 𝒃 𝟎 ] OR 𝐴 = [ 𝒂 𝟎 ] OR 𝐴=[ 𝟎 𝒃 𝑔 ] OR 𝐴=[ ]
𝑐 𝒃 𝟎 𝒃
𝑓 𝑔 𝒄 𝟎 𝟎 𝒄

 In these cases, the eigenvalues are on the diagonal: 𝜆1 = 𝑎, 𝜆2 = 𝑏 and 𝜆3 = 𝑐

̅=0
 To determine eigenvectors using (𝐴 − 𝜆𝐼 )𝐾 ̅ , look out for the following:
𝟎 𝑎 𝑏 𝑎 𝟎 𝑏 𝑎 𝑏 𝟎
(𝐴 − 𝜆𝐼) = [ 𝟎 𝑐 𝑑 ] OR (𝐴 − 𝜆𝐼) = [ 𝑐 𝟎 𝑑 ] OR (𝐴 − 𝜆𝐼)= [ 𝑐 𝑑 𝟎 ]
𝟎 𝑒 𝑓 𝑒 𝟎 𝑓 𝑒 𝑓 𝟎

𝑲𝟏 ∈ ℝ 𝑲𝟐 ∈ ℝ 𝑲𝟑 ∈ ℝ

 To prove that 𝑋̅(𝑡) is a solution of 𝑋̅ ′ = 𝐴𝑋̅ (or 𝑋̅ ′ = 𝐴𝑋̅ + 𝐹̅ ), substitute 𝑋̅(𝑡) and 𝑋̅′(𝑡)
into the equation and check if LHS=RHS
 |𝐴 − 𝜆𝐼 | = 0 : characteristic equation
Complex eigenvalues
̅𝒏 = 𝑲
 The solution is in the form 𝑿 ̅ 𝒏 𝒆𝝀𝒏 𝒕, with 𝝀𝒏 = 𝜶 ± 𝜷𝒊

 Definition: 𝒆𝜶+𝜷𝒊 = 𝒆𝜶 𝒆𝜷𝒊 = 𝒆𝜶 (𝐜𝐨𝐬 𝜷 + 𝒊 𝐬𝐢𝐧 𝜷)

 Complex eigenvalues exist in conjugate pairs.
 If 𝝀 = 𝜶 + 𝜷𝒊 is an eigenvalue, 𝝀̅ = 𝜶 − 𝜷𝒊 is also an eigenvalue
̅ is an eigenvector
̅ is an eigenvector corresponding to 𝝀 = 𝜶 + 𝜷𝒊 , then 𝑲
 If 𝑲
corresponding to 𝝀̅ = 𝜶 − 𝜷𝒊
̅ in order to find two real solutions:
̅ or 𝝀̅ and 𝑲
 Choose to work with either 𝝀 and 𝑲

̅ = 𝒆(𝜶+𝜷𝒊)𝒕 (𝑨 + 𝒊𝑩)
̅ ∗ = 𝒆𝝀𝒕 𝑲
𝑿

̅ ∗ = 𝒆𝜶𝒕 (𝐜𝐨𝐬 𝜷𝒕 + 𝒊 𝐬𝐢𝐧 𝜷𝒕)(𝑨 + 𝒊𝑩)

𝑿

̅ ∗ = 𝒆𝜶𝒕 [(𝑨 𝐜𝐨𝐬 𝜷𝒕 − 𝑩 𝐬𝐢𝐧 𝜷𝒕) + 𝒊(𝑨 𝐬𝐢𝐧 𝜷𝒕 + 𝑩 𝐜𝐨𝐬 𝜷𝒕)]

𝑿

ℝ ℂ

̅ 𝟏 = ℝ (𝑿
𝑿 ̅ ∗) = 𝒆𝜶𝒕 (𝑨 𝐜𝐨𝐬 𝜷𝒕 − 𝑩 𝐬𝐢𝐧 𝜷𝒕)

̅ 𝟐 = ℂ(𝑿
𝑿 ̅ ∗ ) = 𝒆𝜶𝒕 (𝑨 𝐬𝐢𝐧 𝜷𝒕 + 𝑩 𝐜𝐨𝐬 𝜷𝒕)

1.2 Repeated eigenvalue solutions

Definition
An eigenvalue is of multiplicity 𝒎 if it appears 𝒎 times as a root of |𝑨 − 𝝀𝑰| = 𝟎

 An eigenvalue with multiplicity 𝑚 is complete if it has 𝑚 corresponding linearly

independent eigenvectors
 If the eigenvalue is complete:
̅ (𝒕) = 𝒄𝟏 𝒆𝝀𝟏 𝒕 𝑲
𝑿 ̅ 𝟏 + 𝒄𝟐 𝒆𝝀𝟐 𝒕 𝑲
̅ 𝟐 + 𝒄𝟑 𝒆𝝀𝟑 𝒕 𝑲
̅ 𝟑 + ⋯ + 𝒄𝒏 𝒆𝝀𝒏 𝒕 𝑲
̅𝒏
 If the eigenvalue is not complete: (𝜆1 = 𝜆2 = 𝜆3 = ⋯ = 𝜆𝑛 )
𝑿
̅ 𝟏 = 𝒆𝝀𝟏 𝒕 𝑲
̅ ̅ determined using (𝑨 − 𝝀𝑰)𝑲
with 𝑲 ̅=𝟎
̅

𝑿
̅ 𝟐 = 𝒕𝒆𝝀𝟏 𝒕 𝑲
̅ + 𝒆𝝀𝟏 𝒕 𝑷
̅ ̅ determined using (𝑨 − 𝝀𝑰)𝑷
with 𝑷 ̅=𝑲
̅
𝒕𝟐
𝑿
̅𝟑 = ̅ + 𝒕𝒆𝝀𝟏 𝒕 𝑷
𝒆𝝀𝟏 𝒕 𝑲 ̅ + 𝒆𝝀𝟏 𝒕 𝑸
̅ ̅ determined using (𝑨 − 𝝀𝑰)𝑸
with 𝑸 ̅=𝑷
̅
𝟐

̅ (𝒕) = 𝑿
𝑿 ̅𝟏 + 𝑿
̅𝟐 + 𝑿
̅𝟑
1.3 Nonhomogeneous linear systems

To solve the non-homogeneous system: 𝑋̅ ′ = 𝐴𝑋̅ + 𝐹̅

̅ ( 𝒕) = 𝑿
Solution is in the form 𝑿 ̅𝒄 + 𝑿
̅𝒑

̅ 𝒄 is the solution of the associated homogeneous system 𝑋̅ ′ = 𝐴𝑋̅

 𝑿
̅ 𝒑 is the particular solution with its form determined using 𝐹̅ :
 𝑿

• ̅ = (𝒇𝟏 )
𝑭 ̅ 𝒑 = (𝒂)
𝑿
𝒇𝟐 𝒃

• ̅ = 𝒕 (𝒇𝟏 )
𝑭 ̅ 𝒑 = 𝒕 (𝒂) + ( 𝒄 )
𝑿
𝒇𝟐 𝒃 𝒅
polynomials

𝒆
• ̅ = 𝒕𝟐 (𝒇𝟏 )
𝑭 ̅ 𝒑 = 𝒕𝟐 (𝒂) + 𝒕 ( 𝒄 ) + ( )
𝑿
𝒇𝟐 𝒃 𝒅 𝒇
𝒆
• ̅ = 𝒕𝟑 (𝒇𝟏 )
𝑭 ̅ 𝒑 = 𝒕𝟑 (𝒂) + 𝒕𝟐 ( 𝒄 ) + 𝒕 ( ) + (𝒈)
𝑿
𝒇𝟐 𝒃 𝒅 𝒇 𝒉

• ̅ = 𝐬𝐢𝐧 𝒕 (𝒇𝟏 )
𝑭 ̅ 𝒑 = 𝐬𝐢𝐧 𝒕 (𝒂) + 𝐜𝐨𝐬 𝒕 ( 𝒄 )
𝑿
𝒇𝟐 𝒃 𝒅

• ̅ = 𝐜𝐨𝐬 𝒕 (𝒇𝟏 )
𝑭 ̅ 𝒑 = 𝐜𝐨𝐬 𝒕 (𝒂) + 𝐬𝐢𝐧 𝒕 ( 𝒄 )
𝑿
𝒇𝟐 𝒃 𝒅

• ̅ = 𝒆𝜶𝒕 (𝒇𝟏 )
𝑭 ̅ 𝒑 = 𝒆𝜶𝒕 (𝒂)
𝑿
𝒇𝟐 𝒃

̅ 𝒑 with 𝑿
 Always compare 𝑿 ̅ 𝒄 and look out for duplication:
̅ 𝒑 the same
 No duplication: keep 𝑿
̅ 𝒑 with a polynomial of higher degree then highest degree of the
 Duplication: multiply 𝑿
̅𝒄
polynomial in 𝑿
 For combinations of different forms in 𝐹̅ :

• ̅ = [𝒑𝒐𝒍𝒊𝒏𝒐𝒎𝒊𝒂𝒍 × 𝒆𝜶𝒕 ] (𝒇𝟏 )

𝑭
𝒇𝟐
𝒆
̅ 𝒑 = 𝒆𝜶𝒕 [𝒕𝟐 (𝒂) + 𝒕 ( 𝒄 ) + ( )]
𝑿
𝒃 𝒅 𝒇

• ̅ = [𝒑𝒐𝒍𝒊𝒏𝒐𝒎𝒊𝒂𝒍 × (𝐬𝐢𝐧 𝒕 𝒐𝒓 𝐜𝐨𝐬 𝒕) ] (𝒇𝟏 )

𝑭
𝒇𝟐
𝒆 𝒆
̅ 𝒑 = 𝐜𝐨𝐬 𝒕 [𝒕𝟐 (𝒂) + 𝒕 ( 𝒄 ) + ( )] + 𝐬𝐢𝐧 𝒕 [𝒕𝟐 (𝒂) + 𝒕 ( 𝒄 ) + ( )]
𝑿
𝒃 𝒅 𝒇 𝒃 𝒅 𝒇

• ̅ = [ 𝒆𝜶𝒕 × (𝐬𝐢𝐧 𝒕 𝒐𝒓 𝐜𝐨𝐬 𝒕)] (𝒇𝟏 )

𝑭
𝒇𝟐
̅ 𝒑 = 𝒆𝜶𝒕 [𝐬𝐢𝐧 𝒕 (𝒂) + 𝐜𝐨𝐬 𝒕 ( 𝒄 )]
𝑿
𝒃 𝒅
1.4 Second-order linear systems

To solve the system: 𝑋̅ ′′ = 𝐴𝑋̅

̅ (𝒕) = 𝑲
Solution in the form 𝑿 ̅ 𝒆𝝀𝒕

̅ (𝒕) = (𝒂𝟏 𝐜𝐨𝐬 𝝎𝟏 𝒕 + 𝒃𝟏 𝐬𝐢𝐧 𝝎𝟏 𝒕)𝑲

 𝑿 ̅ 𝟏 + ⋯ + (𝒂𝒏 𝐜𝐨𝐬 𝝎𝒏 𝒕 + 𝒃𝒏 𝐬𝐢𝐧 𝝎𝒏 𝒕)𝑲
̅𝒏
with 𝝀 = −𝝎𝟐 , 𝜔>0
 Determine 𝝀 with |𝑨 − 𝝀𝑰| = 𝟎
̅ with (𝑨 − 𝝀𝑰)𝑲
 Determine 𝑲 ̅ =𝟎
̅
Theme 2: Sequences and series of real numbers

2.1 Sequences of real numbers

{𝒂𝒏 } = {𝒂𝟏 , 𝒂𝟐 , … , 𝒂𝒏 }

Using 𝐥𝐢𝐦 𝒂𝒏 to determine convergence:

𝒏→∞

 If 𝐥𝐢𝐦 𝒂𝒏 exists, {𝒂𝒏 } is convergent

𝒏→∞

 If 𝐥𝐢𝐦 𝒂𝒏 does not exists, {𝒂𝒏 } is divergent

𝒏→∞

Geometric sequence: {𝒓𝒏 }

 Convergent if −𝟏 < 𝒓 ≤ 𝟏
 𝐥𝐢𝐦 𝒓𝒏 = 𝟏 if 𝑟=1
𝒏→∞

 𝐥𝐢𝐦 𝒓𝒏 = 𝟎 if |𝑟 | < 1
𝒏→∞

 Divergent if 𝒓 > 𝟏 or 𝒓 ≤ −𝟏
 𝐥𝐢𝐦 𝒓𝒏 = ∞ if 𝒓>𝟏
𝒏→∞

 𝐥𝐢𝐦 𝒓𝒏 does not exist if 𝒓 ≤ −𝟏

𝒏→∞

Depending on the form of 𝑎𝑛 L’Hospital can be used to determine 𝐥𝐢𝐦 𝒂𝒏

𝒏→∞
0
 Determine lim 𝑓(𝑥) using L’Hospital, 𝑎𝑛 = 𝑓(𝑛)
0 𝑥→∞
±∞ 0 𝑝𝑜𝑙𝑖𝑛𝑜𝑚𝑖𝑎𝑙
 Same as [OR for a rational function ( ), divide by 𝑛ℎ𝑖𝑔ℎ𝑒𝑠𝑡 𝑑𝑒𝑔𝑟𝑒𝑒 ]
±∞ 0 𝑝𝑜𝑙𝑖𝑛𝑜𝑚𝑖𝑎𝑙
𝑐𝑜𝑛𝑗𝑢𝑔𝑎𝑡𝑒
 ∞−∞ Factorise [OR multiply with (𝑐𝑜𝑛𝑗𝑢𝑔𝑎𝑡𝑒)]

 0×∞ Rewrite as a fraction and use L’Hospital

 00 (1) lim 𝑓(𝑥) ⇒ lim ln 𝑓(𝑥)
METHOD

𝑥→∞ 𝑥→∞
0
 ∞ (2) Use ln 𝑓(𝑥) characteristics, rewrite as a fraction and use L’Hospital
 1∞ (3) lim ln 𝑓(𝑥) ⇒ lim 𝑓(𝑥) [Remember this step!]
𝑥→∞ 𝑥→∞

L’Hospital’s rule:
𝑔(𝑥) 𝒈(𝒙)
 If 𝑓(𝑥 ) = ℎ(𝑥) , 𝐥𝐢𝐦 𝒇(𝒙) = 𝐥𝐢𝐦 𝒉(𝒙)
𝒙→∞ 𝒙→∞

 Only if 𝑔(𝑥 ) and ℎ(𝑥 ) are differentiable functions

𝒈(𝒙) H 𝒈′(𝒙)
 𝐥𝐢𝐦 𝒉(𝒙) = 𝐥𝐢𝐦 𝒉′(𝒙)
𝒙→∞ 𝒙→∞
Theorem
• Let 𝑓 be a continuous function with 𝐥𝐢𝐦 𝒇(𝒙) = 𝑳
𝒙→∞

• If {𝒂𝒏 } is a sequence so that 𝑎𝑛 = 𝑓(𝑛) for each integer 𝑛, 𝐥𝐢𝐦 𝒂𝒏 = 𝑳

𝒏→∞

Squeeze theorem
Using −𝟏 ≤ 𝐬𝐢𝐧 𝒏 ≤ 𝟏 and −𝟏 ≤ 𝐜𝐨𝐬 𝒏 ≤ 𝟏:

 If 𝒂𝒏 ≤ 𝒃𝒏 ≤ 𝒄𝒏 for 𝑛 ≥ 𝑁 and 𝐥𝐢𝐦 𝒂𝒏 = 𝐥𝐢𝐦 𝒄𝒏 = 𝑳 𝐥𝐢𝐦 𝒃𝒏 = 𝑳

𝒏→∞ 𝒏→∞ 𝒏→∞

Theorem: {𝒂𝒏 } and {𝒃𝒏 } both converge

 𝐥𝐢𝐦 (𝒂𝒏 ± 𝒃𝒏 ) = 𝐥𝐢𝐦 𝒂𝒏 ± 𝐥𝐢𝐦 𝒃𝒏
𝒏→∞ 𝒏→∞ 𝒏→∞

 𝐥𝐢𝐦 (𝒄𝒂𝒏 ) = 𝒄 𝐥𝐢𝐦 𝒂𝒏

𝒏→∞ 𝒏→∞

 𝐥𝐢𝐦 (𝒂𝒏 𝒃𝒏 ) = (𝐥𝐢𝐦 𝒂𝒏 ) (𝐥𝐢𝐦 𝒃𝒏 )

𝒏→∞ 𝒏→∞ 𝒏→∞

𝒂 𝐥𝐢𝐦 𝒂𝒏
 𝐥𝐢𝐦 (𝒃𝒏 ) = 𝒏→∞ with [ lim 𝑏𝑛 ≠ 0]
𝒏→∞ 𝐥𝐢𝐦 𝒃
𝒏 𝑛→∞
𝒏→∞ 𝒏
𝒑
 𝐥𝐢𝐦 𝒂𝒏 𝒑 = (𝐥𝐢𝐦 𝒂𝒏 ) with [ 𝑝 > 0 and 𝑎𝑛 > 0 ]
𝒏→∞ 𝒏→∞

Theorem: |𝑎𝑛 | as an alternating sequence

❖ If lim |𝑎𝑛 | = 0 lim 𝑎𝑛 = 0
𝑛→∞ 𝑛→∞

❖ If lim |𝑎𝑛 | ≠ 0 and {𝒂𝒏 } is an alternating sequence lim 𝑎𝑛 does not exist
𝑛→∞ 𝑛→∞

Theorem: lim 𝑓(𝑎𝑛 )

𝑛→∞

If 𝐥𝐢𝐦 𝒂𝒏 = 𝑳 and 𝑓 is continuous in 𝐿 𝐥𝐢𝐦 𝒇(𝒂𝒏 ) = 𝒇 (𝐥𝐢𝐦 𝒂𝒏 ) = 𝒇(𝑳)

𝒏→∞ 𝒏→∞ 𝒏→∞

 Use this theorem if the sequence is described as 𝒂𝒏 = ⋯ and 𝒂𝒏+𝟏 = ⋯

 𝐥𝐢𝐦 𝒂𝒏 = 𝑳 ⇒ 𝐥𝐢𝐦 𝒂𝒏+𝟏 = 𝑳
𝒏→∞ 𝒏→∞

Definitions:
A sequence {𝒂𝒏 } is:

▪ Increasing if 𝑎𝑛 < 𝑎𝑛+1 for 𝑛 ≥ 1

▪ Decreasing if 𝑎𝑛 > 𝑎𝑛+1 for 𝑛 ≥ 1
▪ Monotonic if it is either increasing or decreasing
3 Methods to determine whether {𝒂𝒏 } is decreasing or increasing:
1) Consider 𝒇′(𝒙) with 𝑓 (𝑛) = 𝑎𝑛 𝒇′ (𝒙) > 𝟎 ⇒ 𝑓 is increasing ⇒ {𝒂𝒏 } is increasing
𝒇′ (𝒙) < 𝟎 ⇒ 𝑓 is decreasing ⇒ {𝒂𝒏 } is decreasing

2) Consider 𝒂𝒏+𝟏 − 𝒂𝒏 𝒂𝒏+𝟏 − 𝒂𝒏 > 0 ⇒ {𝒂𝒏 } is increasing

𝒂𝒏+𝟏 − 𝒂𝒏 < 0 ⇒ {𝒂𝒏 } is decreasing
𝒂𝒏+𝟏 𝒂𝒏+𝟏
3) Consider > 𝟏 and 𝒂𝒏 > 0 ⇒ {𝒂𝒏 } is increasing
𝒂𝒏 𝒂𝒏
𝒂𝒏+𝟏
< 𝟏 and 𝒂𝒏 > 0 ⇒ {𝒂𝒏 } is decreasing
𝒂𝒏

Definitions: bounded
A sequence {𝒂𝒏 } is:

 Bounded above if there exists a number 𝑀 such that 𝑎𝑛 ≤ 𝑀 for 𝑛 ≥ 1

 Bounded below if there exists a number 𝐾 such that 𝑎𝑛 ≥ 𝐾 for 𝑛 ≥ 1
 Bounded if it is bounded above and/or below

A convergent sequence is bounded

 [BUT a bounded sequence is NOT necessarily ]convergent
Every bounded, monotonic sequence is convergent
 An increasing sequence that is bounded above, converges
 A decreasing sequence that is bounded below, converges

2.2 Series of real numbers

∞

∑ 𝒂𝒏 = 𝒂𝟏 + 𝒂𝟐 + 𝒂𝟑 + ⋯
𝒏=𝟏

To determine convergence:
Partial sums: look out for 𝑎𝑛 = 𝑓 (𝑛) = 𝑔(𝑛)
𝑠1 = 𝑎1 Consider the convergence of {𝑠𝑛 }:

𝑠2 = 𝑎1 + 𝑎2  If lim 𝑠𝑛 exists ⇒ ∑∞
𝒏=𝟏 𝒂𝒏 converges
𝑛→∞

𝑠3 = 𝑎1 + 𝑎2 + 𝑎3  If lim 𝑠𝑛 = 𝑠 = ∑∞
𝒏=𝟏 𝒂𝒏
𝑛→∞

Telescopic series
⋮
 If lim 𝑠𝑛 does not exist ⇒ ∑∞
𝒏=𝟏 𝒂𝒏 diverges
𝑛→∞
𝑠𝑛 = 𝑎1 + 𝑎2 + 𝑎3 + ⋯ + 𝑎𝑛
Geometric series
∞ ∞

∑ 𝑎𝑟 𝑛−1 = 𝑎 + 𝑎𝑟 + 𝑎𝑟 2 + ⋯ = ∑ 𝑎𝑟 𝑛
𝑛=1 𝑛=0

Consider 𝑟:
 |𝒓| < 𝟏 series converges
 |𝒓| ≥ 𝟏 series diverges

To determine the sum:

𝑎
• ∑∞
𝑛=1 𝑎𝑟
𝑛−1
= ∑∞ 𝑛
𝑛=0 𝑎𝑟 = 1−𝑟 𝑎 is the first term in the series

Theorem: ∑ 𝑎𝑛 and ∑ 𝑏𝑛 both converge

 ∑∞ ∞
𝑛=1 𝑐𝑎𝑛 = 𝑐 ∑𝑛=1 𝑎𝑛

 ∑∞ ∞ ∞
𝑛=1(𝑎𝑛 ± 𝑏𝑛 ) = ∑𝑛=1 𝑎𝑛 ± ∑𝑛=1 𝑏𝑛

The sum of series

❖ If ∑ 𝑎𝑛 and ∑ 𝑏𝑛 are both convergent, ∑(𝑎𝑛 + 𝑏𝑛 ) converges
❖ If ∑ 𝑎𝑛 converges and ∑ 𝑏𝑛 diverges (or vice versa), ∑(𝑎𝑛 + 𝑏𝑛 ) diverges
❖ If ∑ 𝑎𝑛 and ∑ 𝑏𝑛 are both divergent, ∑(𝑎𝑛 + 𝑏𝑛 ) is either convergent or divergent

Test for divergence:

 If ∑∞
𝑛=1 𝑎𝑛 converges, lim 𝑎𝑛 = 0
𝑛→∞

 If lim 𝑎𝑛 ≠ 0 or lim 𝑎𝑛 does not exist, ∑∞

𝑛=1 𝑎𝑛 diverges
𝑛→∞ 𝑛→∞

 If lim 𝑎𝑛 = 0, nothing can be determined

𝑛→∞

The including or excluding of a finite number of terms does not influence the convergence of a
series

Harmonic series
∞ ∞
1
∑ 𝑎𝑛 = ∑ Harmonic series
𝑥
𝑛=1 𝑛=1
are divergent.

2.3 Series with non-negative terms

Consider ∑∞
𝑛=1 𝑎𝑛 with 𝑎𝑛 > 0
Integral test
 𝒇 must be continuous, positive and decreasing.

∞
𝐥𝐢𝐦 𝒂𝒏 = 𝟎; 𝑎𝑛 > 0 Consider ∫1 𝑓(𝑥 ). 𝑑𝑥 with 𝑓 (𝑛) = 𝑎𝑛 and decreasing
𝒏→∞

∞
 ∫1 𝑓(𝑥). 𝑑𝑥 converges ⟺ ∑∞
𝑛=1 𝑎𝑛 converges
∞
∫1 𝑓 (𝑥 ). 𝑑𝑥 does NOT determine the sum
∞ of the series! The series (and integral) does
 ∫1 𝑓(𝑥). 𝑑𝑥 diverges ⟺ ∑∞
𝑛=1 𝑎𝑛 diverges
not necessarily have to start at 1.
∞ 𝑡
 𝑅𝑒𝑚𝑒𝑚𝑏𝑒𝑟: ∫1 𝑓(𝑥). 𝑑𝑥 = lim ∫1 𝑓(𝑥)𝑑𝑥
𝑡→∞

The 𝑝-series

∞
1 • 𝑝>1 series converges
∑
𝑛𝑝 • 𝑝≤1 series diverges
𝑛=1

The direct comparison test

 𝑎𝑛 > 0 and 𝑏𝑛 > 0

• 𝑎𝑛 ≤ 𝑏𝑛 for all 𝑛 and ∑∞

𝑛=1 𝑏𝑛 converges ∑∞
𝑛=1 𝑎𝑛 converges

• 𝑎𝑛 > 𝑏𝑛 for all 𝑛 and ∑∞

𝑛=1 𝑏𝑛 diverges ∑∞
𝑛=1 𝑎𝑛 diverges

The limit comparison test

𝒂
 𝑎𝑛 > 0 and 𝑏𝑛 > 0 Consider 𝐥𝐢𝐦 𝒃𝒏
𝒏→∞ 𝒏

𝒂𝒏
• 𝐥𝐢𝐦 = 𝒄, 0 < 𝑐 < ∞
𝒏→∞ 𝒃𝒏

 ∑ 𝑏𝑛 converges ⇒ ∑ 𝑎𝑛 converges
 ∑ 𝑏𝑛 diverges ⇒ ∑ 𝑎𝑛 diverges

𝒂
• 𝐥𝐢𝐦 𝒃𝒏 = 𝟎
𝒏→∞ 𝒏

 ∑ 𝑏𝑛 converges ⇒ ∑ 𝑎𝑛 converges
 ∑ 𝑏𝑛 diverges ⇒ nothing can be determined

𝒂
• 𝐥𝐢𝐦 𝒃𝒏 = ∞
𝒏→∞ 𝒏

 ∑ 𝑏𝑛 converges ⇒ nothing can be determined

 ∑ 𝑏𝑛 diverges ⇒ ∑ 𝑎𝑛 diverges
2.4 Absolute and conditional convergence

 Series containing negative and positive terms.

Alternating series
Let 𝑏𝑛 > 0, with {𝑏𝑛 } decreasing and lim 𝑏𝑛 = 0
𝑛⟶∞

 Then ∑∞ 𝑛 ∞
𝑛=1(−1) 𝑏𝑛 and ∑𝑛=1(−1)
𝑛−1
𝑏𝑛 converges

Estimating the error when an alternating series is approximated by its 𝑛𝑡ℎ partial sum.
|𝑹𝒏 | = |𝑺 − 𝑺𝒏 | ≤ |𝒃𝒏+𝟏 |

 𝑺 is the exact sum of the series

 |𝑺 − 𝑺𝒏 | is the absolute value of the error when 𝑺 is approximated using 𝑺𝒏
 |𝒃𝒏+𝟏 | is the absolute value of the first term that is not used.

Absolute and conditional convergence

∑ 𝑎𝑛 is….

• Absolute convergent ∑∞
𝑛=1|𝑎𝑛 | is convergent

• Conditionally convergent ∑∞ ∞
𝑛=1|𝑎𝑛 | is divergent and ∑𝑛=1 𝑎𝑛 is convergent

• Divergent ∑∞ ∞
𝑛=1|𝑎𝑛 | is divergent and ∑𝑛=1 𝑎𝑛 is divergent

 If ∑∞
𝑛=1 𝑎𝑛 is absolute convergent, the series is convergent, but the inverse is not true.

The ratio test

𝑎𝑛+1
Consider lim | |=𝐿
𝑛→∞ 𝑎𝑛

 Use the ratio test when 𝑎𝑛 contains faculties (!) or exponents

• 𝐿<1 ⇒ ∑∞
𝑛=1 𝑎𝑛 is absolute convergent (and therefore, convergent)

• 𝐿>1 ⇒ ∑∞
𝑛=1 𝑎𝑛 is divergent (also true if 𝐿 = ∞)

• 𝐿=1 ⇒ nothing can be determined

The root test
Consider lim 𝑛√|𝑎𝑛 | = 𝐿
𝑛→∞

 Use the root test when 𝑎𝑛 contains 𝑛𝑡ℎ powers

• 𝐿<1 ⇒ ∑∞
𝑛=1 𝑎𝑛 is absolute convergent (and therefore, convergent)

• 𝐿>1 ⇒ ∑∞
𝑛=1 𝑎𝑛 is divergent (also true if 𝐿 = ∞)

• 𝐿=1 ⇒ nothing can be determined

Extra information:

𝑎 𝑛 sin(1⁄𝑛)
• lim (1 + 𝑛) = 𝑒 𝑎 • lim =1
𝑛→∞ 𝑛→∞ 1⁄𝑛

• lim 𝑛√𝑛 = 1 • lim

tan(1⁄𝑛)
=1
𝑛→∞
𝑛→∞ 1⁄𝑛
Theme 3: Power series

3.1 Convergence of a power series

A power series is a series in the form ∑∞ 𝒏 𝟐

𝒏=𝟎 𝒄𝒏 𝒙 = 𝒄𝟎 + 𝒄𝟏 𝒙 + 𝒄𝟐 𝒙 + ⋯

 𝑥 is a variable
 𝑐𝑛′ 𝑠 are constants
 This is called a power series centered at 0 (or about 0,or a power series in 𝑥)

A series in the form ∑∞ 𝒏 𝟐

𝒏=𝟎 𝒄𝒏 (𝒙 − 𝒂) = 𝒄𝟎 + 𝒄𝟏 (𝒙 − 𝒂) + 𝒄𝟐 (𝒙 − 𝒂) + ⋯

 Power series in (𝑥 − 𝑎)
 Power series with central point 𝑎
 Power series about 𝑎

A power series as a function: 𝑓(𝑥 ) = ∑∞

𝑛=0 𝑐𝑛 (𝑥 − 𝑎)
𝑛

 The domain of the function 𝑓 is the set of values of 𝑥 for which the series converges,
- 𝐼 = convergence interval
 For every fixed 𝑥, the power series is a series of constants that can be tested for
convergence.
 𝑅 = radius of convergence
 Every power series is convergent at its centre

The theorem about the existence of a radius of convergence of a power series

For a given power series ∑∞ 𝑛
𝑛=0 𝑐𝑛 (𝑥 − 𝑎) , exactly ONE of the following is true:

• The series ONLY converges for 𝑥 = 𝑎 ⇒ 𝑅 = 0; 𝐼 = {𝑎}

𝑎

divergent divergent
convergent

• The series converges for all 𝑥 ∈ ℝ ⇒ 𝑅 = ∞; 𝐼 = (−∞, ∞)

𝑎

convergent
 • There exists a number 𝑅 > 0, so that the series
 converges for |𝑥 − 𝑎| < 𝑅 and
 diverges if |𝑥 − 𝑎| > 𝑅
 𝑅=𝑅 𝑎−𝑅 𝑎 𝑎+𝑅
[𝑎 − 𝑅, 𝑎 + 𝑅]
(𝑎 − 𝑅, 𝑎 + 𝑅)
 𝐼={ }
(𝑎 − 𝑅, 𝑎 + 𝑅] divergent divergent
[𝑎 − 𝑅, 𝑎 + 𝑅) convergent

3.2 Representation of functions as power series

𝟏
∑∞ 𝒏
𝒏=𝟎 𝒙 = 𝟏−𝒙 if |𝑥 | < 1

 Used to write simple rational functions as power series

 Functions can be integrated or differentiated to get it into the correct form

Differentiation and integration of power series

Let the radius of convergence of 𝑓(𝑥 ) = ∑∞ 𝑛
𝑛=0 𝑐𝑛 (𝑥 − 𝑎) is 𝑅 > 0

 Then 𝑓 is continuous on (𝑎 − 𝑅, 𝑎 + 𝑅) and is therefore also differentiable and integrable

on (𝑎 − 𝑅, 𝑎 + 𝑅)

Differentiation
∞ ∞ ∞
′(
𝑑 𝑑
𝑓 𝑥) = [∑ 𝑐𝑛 (𝑥 − 𝑎)𝑛 ] = ∑ [𝑐 (𝑥 − 𝑎)𝑛 ] = ∑ 𝑛𝑐𝑛 (𝑥 − 𝑎)𝑛−1
𝑑𝑥 𝑑𝑥 𝑛
𝑛=0 𝑛=0 𝑛=0

∴ 𝒇′ (𝒙) = ∑ 𝒏𝒄𝒏 (𝒙 − 𝒂)𝒏−𝟏

𝒏=𝟏

Integration
∞ ∞ ∞
𝑐𝑛 (𝑥 − 𝑎)𝑛+1
∫ 𝑓 (𝑥 ). 𝑑𝑥 = ∫ [∑ 𝑐𝑛 (𝑥 − 𝑎 )𝑛 ] . 𝑑𝑥 = ∑ ∫[𝑐𝑛 (𝑥 − 𝑎 )𝑛 ] . 𝑑𝑥 = 𝑐 + ∑
𝑛+1
𝑛=0 𝑛=0 𝑛=0

∞
𝒄𝒏 (𝒙 − 𝒂)𝒏+𝟏
∴ ∫ 𝒇(𝒙). 𝒅𝒙 = 𝒄 + ∑
𝒏+𝟏
𝒏=𝟎

 Remember to only differentiate and integrate 𝑥, do not integrate or differentiate in terms of 𝑛

 When differentiating or integrating a power series, the radius of convergence DOES NOT change;
but the convergence interval might change.

3.3 Taylor polynomials and Taylor series

𝒇 (𝒙 ) = ∑ ∞
𝒏=𝟎 𝒄𝒏 (𝒙 − 𝒂)
𝒏
for |𝑥 − 𝑎| < 𝑅

Theorem: Taylor series

If 𝑓 has a power series about 𝑎 (i.e. 𝑓 (𝑥 ) = ∑∞ 𝑛
𝑛=0 𝑐𝑛 (𝑥 − 𝑎) for all |𝑥 − 𝑎| < 𝑅):

𝒇(𝒏) (𝒂)
 𝒄𝒏 = , 𝑛≥0
𝒏!

Taylor series
∞ ∞
𝑓 (𝑛) (𝑎)
𝑓 (𝑥 ) = ∑ 𝑐𝑛 (𝑥 − 𝑎 )𝑛 ⇒ 𝑓 (𝑥 ) = ∑ (𝑥 − 𝑎 ) 𝑛
𝑛!
𝑛=0 𝑛=0

Maclaurin series (𝑎 = 0)
∞ ∞
𝑓 (𝑛) (𝑎) 𝑓 (𝑛)(0) 𝑛
𝑓 (𝑥 ) = ∑ (𝑥 − 𝑎 ) 𝑛 ⇒ 𝑓 (𝑥 ) = 𝑓 (𝑥 ) = ∑ 𝑥
𝑛! 𝑛!
𝑛=0 𝑛=0

To determine the Taylor or Maclaurin series of a function:

Given: 𝑓 (𝑥 ) = ⋯

1) Determine 𝑓 (𝑛) (𝑥) for 𝑛 = 1,2,3,4 … until a pattern becomes evident.

 Do not simplify!
2) Substitute with power series form
 It might be necessary to make adaptions if it is clear from 𝑓 (𝑛)(𝑥) that it is an alternating
series or that it is only valid for even/odd terms.

When does the Taylor series of a function converge to the function?

𝑓(𝑛) (𝑎)
(When is 𝑓(𝑥) truly determined by ∑∞
𝑛=0 (𝑥 − 𝑎)𝑛 ?)
𝑛!

𝑓(𝑛) (𝑎)
 𝑇𝑛 (𝑥 ) = ∑∞
𝑛=0 (𝑥 − 𝑎 ) 𝑛 𝑛𝑡ℎ degree Taylor-polynomial of 𝑓 about 𝑎
𝑛!

 𝑇𝑛 the 𝑛𝑡ℎ partial sum of the Taylor series of 𝑓 about 𝑎

gives an approximation of 𝑓 close to 𝑎
 Remainder: 𝑅𝑛 (𝑥 ) = 𝑓 (𝑥 ) − 𝑇𝑛 (𝑥 )
  The sequence of partial sums {𝑇𝑛 } converges to 𝑓(𝑥 ) for the values of 𝑥 for
which lim 𝑅𝑛 (𝑥 ) = 0
𝑛⟶∞

𝐥𝐢𝐦 𝑹𝒏 (𝒙) = 𝟎 ⇒ 𝐥𝐢𝐦 [𝒇(𝒙) − 𝑻𝒏 (𝒙)] = 𝟎

𝒏⟶∞ 𝒏⟶∞

∴ lim 𝑇𝑛 (𝑥 ) = 𝑓 (𝑥 )
𝑛⟶∞

𝑓(𝑛) (𝑎)
∴ 𝑓 (𝑥 ) = ∑ ∞
𝑛=0 (𝑥 − 𝑎)𝑛 for all values of 𝑥 for which lim 𝑅𝑛 (𝑥 ) = 0
𝑛! 𝑛⟶∞

Theorem
If 𝑓 (𝑥 ) = 𝑇𝑛 (𝑥 ) + 𝑅𝑛 (𝑥 ), where 𝑇𝑛 (𝑥) is the 𝑛𝑡ℎ degree Taylor-polynomial of 𝑓 about 𝑎 and
𝑓(𝑛) (𝑎)
lim 𝑅𝑛 (𝑥 ) = 0 for |𝑥 − 𝑎| < 𝑅, then 𝑓 (𝑥 ) = ∑∞
𝑛=0 (𝑥 − 𝑎)𝑛 is valid.
𝑛⟶∞ 𝑛!

Taylor’s theorem
If 𝑓 has a (𝑛 + 1)𝑡ℎ differential at every point in an open interval 𝐼 and 𝑎 ∈ 𝐼, then there is a
number 𝑐 between 𝑥 and 𝑎 so that:

𝑓(𝑛+1) (𝑐)
|𝑅𝑛 (𝑥)| = | (𝑥 − 𝑎)𝑛+1 | with 𝑐 between 𝑥 and 𝑎
(𝑛+1)!

𝑓 ′′ (𝑎) 𝑓 (𝑛) (𝑎) 𝑓 (𝑛+1) (𝑐)

𝑓(𝑥 ) = 𝑓(𝑎) + 𝑓 ′ (𝑎)(𝑥 − 𝑎) + (𝑥 − 𝑎)2 + ⋯ + (𝑥 − 𝑎 )𝑛 + (𝑥 − 𝑎)𝑛+1
2! 𝑛! (𝑛 + 1)!

𝑇𝑛 (𝑥) 𝑅𝑛 (𝑥)
Standard limit:
𝑥𝑛
lim = 0 for all 𝑥 ∈ ℝ
𝑛→∞ 𝑛!

𝑥𝑛
 The series ∑∞ ∞
𝑛=0 𝑎𝑛 ⇒ ∑𝑛=0 𝑛! converges for all 𝑥 ∈ ℝ ∴ lim 𝑎𝑛 = 0
𝑛→∞

The Binomial series

Maclaurin series for 𝑓 (𝑥 ) = (1 + 𝑥)𝑘 , 𝑘 ∈ ℝ

𝒌 𝒏
(𝟏 + 𝒙)𝒌 = ∑∞
𝒏=𝟎(𝒏)𝒙 , with |𝑥 | < 1

𝑘(𝑘−1)(𝑘−2)(𝑘−3)…..(𝑘−𝑛+1)
 (𝑘𝑛) = (number of terms = 𝑛)
𝑛!

 (𝑘0) = 1
 𝑘>0 definite number of terms
 𝑘<0 indefinite number of terms
3.4 Power series solution of differential equations

Use the power series 𝑦 = ∑∞ 𝑛

𝑛=0 𝑐𝑛 𝑥 to solve a DE and to determine the coefficients

𝑐0 , 𝑐1 , 𝑐2, 𝑐3 , 𝑐4, … , 𝑐𝑛

Differentiation theorem
𝑓 (𝑥 ) = ∑ ∞
𝑛=0 𝑐𝑛 𝑥
𝑛
converges on an open interval 𝐼

⇒ 𝑓 ′ (𝑥 ) = ∑ ∞
𝑛=0 𝑛𝑐𝑛 𝑥
𝑛−1
for all 𝑥 ∈ 𝐼

Identity principle
If ∑∞ 𝒏 ∞
𝒏=𝟎 𝒂𝒏 𝒙 = ∑𝒏=𝟎 𝒃𝒏 𝒙
𝒏
for all 𝑥 in an open interval 𝐼

 𝒂𝒏 = 𝒃𝒏 for all 𝑛 ≥ 0

If ∑∞ 𝒏
𝒏=𝟎 𝒂𝒏 𝒙 = 𝟎 for all 𝑥 ∈ 𝐼

 𝒂𝒏 = 𝟎 for all 𝑛 ≥ 0

Radius of convergence
• A solution in the form 𝑦 = ∑∞
𝑛=0 𝑐𝑛 𝑥
𝑛
⇒ 𝑎𝑛 = 𝑐𝑛 𝑥 𝑛
• Use the ratio test to determine the radius of convergence:
𝑎𝑛+1
 Consider lim | |=𝐿
𝑛→∞ 𝑎𝑛

- 𝐿<1 series converges

- 𝐿>1 series diverges
- 𝐿=1 nothing can be determined

Shifting the index of summation

Consider ∑∞
𝒏=𝟏 𝒏𝒄𝒏 𝒙
𝒏−𝟏

 Let 𝑘 = 𝑛 − 1 ⇒ 𝑛 = 𝑘+1 (if 𝑛 = 1 ⇒ 𝑘 = 0)

 ∑∞
𝑛=1 𝑛𝑐𝑛 𝑥
𝑛−1
⇒ ∑∞ 𝒌
𝒌=𝟎(𝒌 + 𝟏)(𝒄𝒌+𝟏 𝒙 )

• If the index (everywhere 𝑛 is written) increases by 𝑘, you have to decrease the starting point of
the summation by 𝑘
• If the index decreases by 𝑘, you have to increase the starting point of the summation by 𝑘
Method to solve a Differential Equation (DE) using a power series:
1) Let 𝑦 = ∑∞
𝑛=0 𝑐𝑛 𝑥
𝑛

2) Differentiate 𝑦
- 1st order DE: determine 𝑦′
- 2nd order DE: determine 𝑦′′
3) Substitute into DE
4) Shift indexes to have 𝑥 𝑛 everywhere in the equation
5) Get the summation indexes to have the same starting point
- Write out some of the terms (usually the first 1 or 2 terms)
6) Compare coefficients
7) Determine the recurrence relation using step 6
For which values
8) Determine 𝑐𝑛 using the recurrence relation
of 𝑛 is this valid?
9) Substitute back into solution: 𝑦 = ∑∞
𝑛=0 𝑐𝑛 𝑥
𝑛

10) Determine the radius of convergence (𝑅) using the ratio test

Ordinary or singular points

A function is analytical at 𝑥 = 𝑎 if 𝑓 has a power series 𝒇(𝒙) = ∑∞ 𝒏
𝒏=𝟎 𝒄𝒏 (𝒙 − 𝒂) for

|𝑥 − 𝑎| < 𝑅, 𝑅 > 0

 𝑠𝑖𝑛 and 𝑐𝑜𝑠 are analytical everywhere

 𝑒 𝑥 is analytical everywhere
 Polynomials are analytical everywhere
𝑝(𝑥)
 Rational functions (𝑞(𝑥)) are analytical everywhere where 𝑞(𝑥) ≠ 0

Use this to determine the minimum radius of convergence of a power series and to
see if the 2nd order DE has two linearly independent solutions.

Consider 𝑨(𝒙)𝒚′′ + 𝑩(𝒙)𝒚′ + 𝑪(𝒙)𝒚 = 𝟎 with 𝐴, 𝐵 and 𝐶 analytical functions

𝑩(𝒙) 𝑪( 𝒙 )
 Get 𝑦 ′′ alone by dividing with 𝐴(𝑥): 𝒚′′ + 𝑨(𝒙) 𝒚′ + 𝑨(𝒙) 𝒚 = 𝟎
𝑩(𝒙) 𝑪(𝒙 )
 Let 𝑃(𝑥 ) = 𝑨(𝒙) and 𝑄 (𝑥 ) = 𝑨 ( 𝒙 )

 The point 𝑥 = 𝑎 is an ordinary point if 𝑃 (𝑥 ) and 𝑄(𝑥 ) are analytical at 𝑎;

otherwise point 𝑥 = 𝑎 is a singular point
Theorem: ordinary points
If 𝑃(𝑥 ) and 𝑄(𝑥 ) are analytical at 𝑥 = 𝑎, the equation will have two linearly dependent
solutions: each in the form 𝑦(𝑥) = ∑∞
𝑛=0 𝑐𝑛 (𝑥 − 𝑎)
𝑛

Minimum radius of convergence

The distance from 𝑎 to the nearest singular point.
Theme 4: Fourier series

4.1 General Fourier series and convergence

Fourier series
• The inner product of two functions 𝑓1 and 𝑓2 on an open interval [𝑎, 𝑏] is the number
𝑏
(𝑓1, 𝑓2 ) = ∫𝑎 𝑓1 (𝑥 )𝑓2 (𝑥 ) 𝑑𝑥

• Two functions 𝑓1 and 𝑓2 is orthogonal on an open interval [𝑎, 𝑏] if

𝑏
(𝑓1 , 𝑓2 ) = ∫𝑎 𝑓1 (𝑥 )𝑓2(𝑥 ) 𝑑𝑥 = 0

• A set of real functions, {𝜙0 (𝑥 ), 𝜙1 (𝑥 ), 𝜙2 (𝑥 ), … } is orthogonal on an open interval [𝑎, 𝑏]

𝑏
if (𝜙𝑚 , 𝜙𝑛 ) = ∫𝑎 𝜙𝑚 (𝑥 )𝜙𝑛 (𝑥 ) 𝑑𝑥 = 0 with 𝑚 ≠ 𝑛

Consider the set of trigonometric functions:

𝜋 2𝜋 3𝜋 𝜋 2𝜋 3𝜋
{1, sin , sin , sin … , cos , cos , cos …}
𝐿 𝐿 𝐿 𝐿 𝐿 𝐿

 For integers 𝑚 and 𝑛:

𝑝 𝑚𝜋𝑥 𝑛𝜋𝑥
∫−𝑝 (cos 𝑝
) (cos 𝑝
) 𝑑𝑥 = 0 if 𝑚 ≠ 𝑛

𝑝 if 𝑚 = 𝑛
𝑝 𝑚𝜋𝑥 𝑛𝜋𝑥
∫−𝑝 (sin 𝑝
) (sin 𝑝
) 𝑑𝑥 = 0 if 𝑚 ≠ 𝑛

𝑝 if 𝑚 = 𝑛

𝑝 𝑚𝜋𝑥 𝑛𝜋𝑥
∫−𝑝 (sin 𝑝
) (cos 𝑝
) 𝑑𝑥 = 0

𝑝 𝑛𝜋𝑥
∫−𝑝(1) (sin 𝑝
) 𝑑𝑥 = 0 for all 𝑚, 𝑛

𝑝 𝑛𝜋𝑥
∫−𝑝(1) (cos 𝑝
) 𝑑𝑥 = 0

𝜋 2𝜋 3𝜋 𝜋 2𝜋 3𝜋
The set {1, sin 𝐿 , sin 𝐿
, sin 𝐿 … , cos 𝐿 , cos 𝐿 , cos 𝐿 …} is orthogonal on [−𝑝, 𝑝]
 ∞
𝒂𝟎 𝒏𝝅𝒙 𝒏𝝅𝒙
𝒇(𝒙) = + ∑ 𝒂𝒏 𝐜𝐨𝐬 + 𝒃𝒏 𝐬𝐢𝐧
𝟐 𝒑 𝒑
𝒏=𝟏

Fourier coefficients:
1 𝑝
• 𝑎0 = ∫ [𝑓 (𝑥 )]𝑑𝑥
𝑝 −𝑝
1 𝑝 𝑛𝜋𝑥
• 𝑎𝑛 = 𝑝 ∫−𝑝[𝑓 (𝑥 )] [cos ] 𝑑𝑥 Substitute into Fourier series equation
𝑝

1 𝑝 𝑛𝜋𝑥
• 𝑏𝑛 = 𝑝 ∫−𝑝[𝑓(𝑥)] [sin ] 𝑑𝑥
𝑝

A function is piecewise continuous on a closed interval [𝑎, 𝑏] if:

 a finite number of points 𝑥1 < 𝑥2 < ⋯ < 𝑥𝑛 exists on [𝑎, 𝑏], where 𝑓 has finite
discontinuities at these points
 𝑓 continuous on every open interval (𝑥𝑘 , 𝑥𝑘+1 )

Limits of 𝑓(𝑥 + ) and 𝑓 (𝑥 − ) exists and is finite for 𝑎 < 𝑥 < 𝑏

Theorem: Convergence
Let 𝑓 and 𝑓’ be piecewise continuous on the interval (−𝐿, 𝐿) that is, let 𝑓 and 𝑓’ be
continuous except at a finite number of points in the interval and have only finite
discontinuities at these points. Then the Fourier series of 𝑓 on the interval converges to 𝑓(𝑥)
at every point of continuity. At a point of discontinuity the Fourier series converges to the
𝑓(𝑥 + )+𝑓(𝑥 − )
average where 𝑓(𝑥 −) and 𝑓(𝑥 +) denote the limit of 𝑓 at 𝑥 from the left and
2

from the right, respectively.

 The Fourier series of 𝑓 (for all 𝑥) on the open interval (−𝑝, 𝑝) converges to:
• The value of 𝑓(𝑥) in every point where 𝑓 is continuous
1
• The value 2 [𝑓(𝑥 +) + 𝑓(𝑥 − )] in every point where 𝑓 is discontinuous

- (The average of the left- and right limits of 𝑓 at 𝑥)

Sketch 𝑓, as well as the periodic extension and clearly indicate where 𝑓 is continuous and
discontinuous. Use the theorem to indicate how 𝑓 converges.

Considering that all Fourier series has a period of 2𝑝, the Fourier series not only represents 𝑓,
but also provides the periodic extension of 𝑓 (extended function)with period 2𝑝.
Using the Fourier series to determine the sum
• Sketch the extended function to determine how 𝑓 converges
• What should 𝑥 be to determine the sum? (Manipulation)

4.2 Fourier sine and cosine series

Even and odd functions

 𝑓 is even if 𝑓 (−𝑥 ) = 𝑓 (𝑥 )
 𝑓 is odd if 𝑓 (−𝑥 ) = −𝑓 (𝑥 )

Characteristics:
𝑎 𝑎
 If 𝑓 is even ∫−𝑎 𝑓(𝑥)𝑑𝑥 = 2 ∫0 𝑓(𝑥)𝑑𝑥
𝑎
 If 𝑓 is odd ∫−𝑎 𝑓(𝑥)𝑑𝑥 = 0
 The product of two even functions is even 𝑒𝑣𝑒𝑛 × 𝑒𝑣𝑒𝑛 = 𝑒𝑣𝑒𝑛
 The product of two odd functions is even 𝑜𝑑𝑑 × 𝑒𝑣𝑒𝑛 = 𝑒𝑣𝑒𝑛
 The product of an even and odd function is uneven 𝑒𝑣𝑒𝑛 × 𝑜𝑑𝑑 = 𝑜𝑑𝑑
 The sum (or difference) of two even functions is even 𝑒𝑣𝑒𝑛 ± 𝑒𝑣𝑒𝑛 = 𝑒𝑣𝑒𝑛
 The sum (or difference) of two odd functions is odd 𝑜𝑑𝑑 ± 𝑜𝑑𝑑 = 𝑜𝑑𝑑

Make use of even and odd characteristics to determine 𝑎0 , 𝑎𝑛 and 𝑏𝑛

Sine and cosine series

𝑛𝜋𝑥
 sin odd
𝑝
𝑛𝜋𝑥
 cos even
𝑝

𝒂𝟎 𝒏𝝅𝒙 𝒏𝝅𝒙
Fourier series: 𝒇(𝒙) = + ∑∞
𝒏=𝟏 𝒂𝒏 𝐜𝐨𝐬 + 𝒃𝒏 𝐬𝐢𝐧
𝟐 𝒑 𝒑
1 𝑝
• 𝑎0 = ∫ [𝑓 (𝑥 )]𝑑𝑥
𝑝 −𝑝
1 𝑝 𝑛𝜋𝑥
• 𝑎𝑛 = ∫−𝑝[𝑓 (𝑥 )] [cos ] 𝑑𝑥
𝑝 𝑝
1 𝑝 𝑛𝜋𝑥
• 𝑏𝑛 = 𝑝 ∫−𝑝[𝑓(𝑥)] [sin ] 𝑑𝑥
𝑝
If 𝑓 is an odd function defined on (−𝑝, 𝑝):
1 𝑝
• 𝑎0 = ∫ [𝑓 (𝑥 )]𝑑𝑥 =0 [odd]
𝑝 −𝑝
1 𝑝 𝑛𝜋𝑥
• 𝑎𝑛 = 𝑝 ∫−𝑝[𝑓 (𝑥 )] [cos ] 𝑑𝑥 = 0 [odd x even = odd]
𝑝
1 𝑝 𝑛𝜋𝑥 2 𝑝 𝑛𝜋𝑥
• 𝑏𝑛 = 𝑝 ∫−𝑝[𝑓(𝑥)] [sin ] 𝑑𝑥 = ∫0 [𝑓(𝑥)] [sin ] 𝑑𝑥 [odd x odd = even]
𝑝 𝑝 𝑝

𝒏𝝅𝒙
Fourier sine series: ∑∞
𝒏=𝟏 𝒃𝒏 𝐬𝐢𝐧 𝒑

If 𝑓 is an even function defined on (−𝑝, 𝑝):

1 𝑝 2 𝑝
• 𝑎0 = ∫ [𝑓 (𝑥 )]𝑑𝑥 = 𝑝 ∫0 [𝑓 (𝑥 )]𝑑𝑥 [odd]
𝑝 −𝑝
1 𝑝 𝑛𝜋𝑥 2 𝑝 𝑛𝜋𝑥
• 𝑎𝑛 = 𝑝 ∫−𝑝[𝑓 (𝑥 )] [cos 𝑝
] 𝑑𝑥 = ∫0 [𝑓(𝑥)] [sin
𝑝 𝑝
] 𝑑𝑥 [odd x even = odd]
1 𝑝 𝑛𝜋𝑥
• 𝑏𝑛 = ∫−𝑝[𝑓(𝑥)] [sin ] 𝑑𝑥 = 0 [odd x odd = even]
𝑝 𝑝

𝒂𝟎 𝒏𝝅𝒙
Fourier cosine series: + ∑∞
𝒏=𝟏 𝒂𝒏 𝐜𝐨𝐬
𝟐 𝒑

Extensions: 𝑓 defined on (0, 𝐿)

To show 𝑓 as a trigonometric series:

Sine series
→ Extend 𝑓 on (−𝐿, 𝐿) in such a manner that the new function is odd.
→ Extend 𝑓 to ℝ as a 2𝐿-periodic function
𝒏𝝅𝒙 2 𝐿 𝑛𝜋𝑥
 Fourier sine series: ∑∞
𝒏=𝟏 𝒃𝒏 𝐬𝐢𝐧 with 𝑏𝑛 = 𝐿 ∫0 [𝑓(𝑥)] [sin ] 𝑑𝑥
𝑳 𝐿

Cosine series
→ Extend 𝑓 on (−𝐿, 𝐿) in such a manner that the new function is even.
→ Extend 𝑓 to ℝ as a 2𝐿-periodic function
𝒂𝟎 𝒏𝝅𝒙
 Fourier cosine series: + ∑∞
𝒏=𝟏 𝒂𝒏 𝐜𝐨𝐬
𝟐 𝑳
2 𝐿
with 𝑎0 = 𝐿 ∫0 [𝑓 (𝑥 )]𝑑𝑥
2 𝐿 𝑛𝜋𝑥
𝑎𝑛 = 𝐿 ∫0 [𝑓(𝑥)] [sin ] 𝑑𝑥
𝐿
Fourier series
→ Extend 𝑓 by sketching both 𝑓 and 𝑓(𝑥 + 𝐿).
𝐿
→ Extend 𝑓 to ℝ as an 𝐿-periodic function (2𝑝 = 𝐿 ⇒ 𝑝 = 2)
𝒂𝟎 𝟐𝒏𝝅𝒙 𝟐𝒏𝝅𝒙
 Fourier cosine series: 𝒇(𝒙) =
𝟐
+ ∑∞
𝒏=𝟏 𝒂𝒏 𝐜𝐨𝐬 𝑳
+ 𝒃𝒏 𝐬𝐢𝐧
𝑳
:

2 𝐿
with 𝑎0 = ∫ [𝑓(𝑥)]𝑑𝑥
𝐿 0

2 𝐿 2𝑛𝜋𝑥
𝑎𝑛 = ∫0 [𝑓(𝑥)] [cos ] 𝑑𝑥
𝐿 𝐿

2 𝐿 2𝑛𝜋𝑥
𝑏𝑛 = ∫0 [𝑓(𝑥)] [sin ] 𝑑𝑥
𝐿 𝐿

Using Fourier series to solve endpoint value problems:

Endpoint value problem: 𝒂𝒙′′ + 𝒄𝒙 = 𝒇(𝒕) , 0<𝑡<𝐿

with 𝑥 (0) = 𝑥 (𝐿) = 0 or 𝑥′(0) = 𝑥′(𝐿) = 0

Sine series solution Cosine series solution

𝒏𝝅𝒙 𝒂𝟎 𝒏𝝅𝒙
𝑥 (𝑡 ) = ∑ ∞
𝒏=𝟏 𝒃𝒏 𝐬𝐢𝐧 𝒙(𝒕) = + ∑∞
𝒏=𝟏 𝒂𝒏 𝐜𝐨𝐬
𝑳 𝟐 𝑳

 Determine 𝑥′(𝑡) and 𝑥′′(𝑡) and substitute into the DE to determine 𝑎0 , 𝑎𝑛 and 𝑏𝑛
- Compare coefficients
Theme 5: Partial differential equations

5.1 Eigenvalue problems

2nd order DE’s

• 𝑎𝑦 ′′ + 𝑏𝑦 ′ + 𝑐𝑦 = 0
• Characteristic equation (CE): 𝑎𝑚2 + 𝑏𝑚 + 𝑐 = 0
−𝑏±√𝑏2 −4𝑎𝑐
 𝑚= 2𝑎

Consider the discriminant: 𝒃𝟐 − 𝟒𝒂𝒄

 If 𝒃𝟐 − 𝟒𝒂𝒄 > 𝟎 CE has two distinct real roots 𝑚1 and 𝑚2
the general solution is 𝒚(𝒙) = 𝒄𝟏 𝒆𝒎𝟏 𝒙 + 𝒄𝟐 𝒆𝒎𝟐 𝒙
 If 𝒃𝟐 − 𝟒𝒂𝒄 = 𝟎 CE has a repeated root 𝑚1 = 𝑚2
the general solution is 𝒚(𝒙) = 𝒄𝟏 𝒆𝒎𝒙 + 𝒄𝟐 𝒙𝒆𝒎𝒙
 If 𝒃𝟐 − 𝟒𝒂𝒄 < 𝟎 CE has complex (conjugate) roots 𝑚 = 𝑎 ± 𝑖𝑏, 𝑏 ≠ 0
the general solution is 𝒚(𝒙) = 𝒆𝒂𝒙 (𝒄𝟏 𝐜𝐨𝐬 𝒃𝒙 + 𝒄𝟐 𝐬𝐢𝐧 𝒃𝒙)

Eigenvalues and eigenvectors

𝒚′′ + 𝝀𝒚 = 𝟎, 𝒚(𝟎) = 𝒚(𝑳) = 𝟎, (𝑳 > 𝟎)

 𝝀 is a parameter
 An endpoint value problem with a parameter = eigenvalue problem
o Eigenvalue = value of 𝝀 for which 𝒚′′ + 𝝀𝒚 = 𝟎 has a non-trivial solution
o Eigenvector = non-trivial solution associated with eigenvalue 𝝀
 Any multiple of an eigenvector is also an eigenvector

Method:
1) Use the CE to find the discriminant: ∇ = 𝒃𝟐 − 𝟒𝒂𝒄
2) Consider ∇ = 0; ∇ < 0 [∇ = −𝛼 2 , 𝛼 > 0]; ∇ > 0 [∇ = 𝛼 2 , 𝛼 > 0]
3) Find the eigenvalues (𝜆𝑛 = ⋯) and eigenvectors (𝑦𝑛 = ⋯)
5.2 The heat equation

Consider the heat transfer by conduction in a rod:

• Uniform cross section with area 𝐴
• Homogonous material
• Temperature of material remains constant
• Rod is isolated – no heat can escape other than at the ends of the rod (heat only moves
in the direction of 𝑥)

Boundary value problem

• 𝑢𝑡 = 𝑘𝑢𝑥𝑥 heat equation [𝑢𝑡 must stand alone on the LHS]
• 𝑢(0, 𝑡) = 𝑢(𝐿, 𝑡) = 0 boundary values [endpoint temperature = 0°C]
• 𝑢(𝑥, 0) = 𝑓(𝑥) initial value [initial temperature distribution]

Length of rod = 𝐿 (on 𝑥-axis: from 𝑥 = 0; to 𝑥 = 𝐿)

 If 𝐿 is split into two (or more) rods, draw a picture to visualise the new problem

𝑢(𝑥, 𝑡) = the temperature at point 𝑥 at time 𝑡 (what we want to determine)

Three variations of the boundary value problem

1) 𝑢𝑥 (0, 𝑡) = 𝑢𝑥 (𝐿, 𝑡) = 0 end points insulated
2) 𝑢(0, 𝑡) = 𝑇1 ; 𝑢(𝐿, 𝑡) = 𝑇2 end point at different temperatures
3) Combinations both

End points insulated:

• 𝒖𝒕 = 𝒌𝒖𝒙𝒙
• 𝒖𝒙 (𝟎, 𝒕) = 𝒖𝒙 (𝑳, 𝒕) = 𝟎
• 𝒖(𝒙, 𝟎) = 𝒇(𝒙)

On formula sheet:
−𝒏𝟐 𝝅𝟐 𝒌𝒕⁄ 𝒏𝝅𝒙
𝒖(𝒙, 𝒕) = 𝒄𝟎 + ∑∞
𝒏=𝟏[𝒄𝒏 ] [𝒆 𝑳𝟐 ] [𝐜𝐨𝐬 ]
𝑳
With
1 𝐿
• 𝑐0 = ∫ [𝑓(𝑥)]𝑑𝑥
𝐿 0
2 𝐿 𝑛𝜋𝑥
• 𝑐𝑛 = 𝐿 ∫0 [𝑓 (𝑥 )] [cos 𝐿
] 𝑑𝑥

Tips:
 cos 𝑛𝜋 − 1 ⇒ (−1)𝑛 − 1 → split into even and odd
 Measure time in seconds
 If 𝑓(𝑥) is in the form cos(… 𝑥), do not integrate at 𝑐𝑛 , use 𝑢(𝑥, 0) = ⋯ and compare coefficients
End points at different temperatures:
• 𝒖𝒕 = 𝒌𝒖𝒙𝒙
• 𝒖𝒙 (𝟎, 𝒕) = 𝑻𝟏 ; 𝒖𝒙 (𝑳, 𝒕) = 𝑻𝟐
• 𝒖(𝒙, 𝟎) = 𝒇(𝒙)

On formula sheet:
−𝒏𝟐 𝝅𝟐 𝒌𝒕⁄ 𝒏𝝅𝒙
𝒖(𝒙, 𝒕) = ∑∞
𝒏=𝟏[𝒄𝒏 ] [𝒆 𝑳𝟐 ] [𝐬𝐢𝐧 ]
𝑳
With
• 2 𝐿
𝑐𝑛 = ∫0 [𝑓(𝑥)] [sin
𝐿
𝑛𝜋𝑥
𝐿
] 𝑑𝑥

Tips:
 Use transformations to change the problem back to the standard heat transfer problem
 Unless a specific transformation is specified, use the separation of variables
- 𝑢(𝑥, 𝑡) = 𝑋(𝑥)𝑇(𝑡)

This can be done for any problem that is not in the standard form.

Solving the standard problem:

• 𝒖𝒕 = 𝒌𝒖𝒙𝒙
• 𝒖(𝟎, 𝒕) = 𝒖(𝑳, 𝒕) = 𝟎
• 𝒖(𝒙, 𝟎) = 𝒇(𝒙)

On formula sheet:
−𝒏𝟐 𝝅𝟐 𝒌𝒕⁄ 𝒏𝝅𝒙
𝒖(𝒙, 𝒕) = ∑∞
𝒏=𝟏[𝒄𝒏 ] [𝒆 𝑳𝟐 ] [𝐬𝐢𝐧 ]
𝑳
With
• 2 𝐿
𝑐𝑛 = ∫0 [𝑓(𝑥)] [sin
𝐿
𝑛𝜋𝑥
𝐿
] 𝑑𝑥

Tips:
 If 𝑓(𝑥) is in the form sin(… 𝑥), do not integrate at 𝑐𝑛 , use 𝑢(𝑥, 0) = ⋯ and compare coefficients

5.3 The wave equation

Consider the displacement of a vibrating string fastened between points 𝑥 = 0 and 𝑥 = 𝐿

 String is in the 𝑥𝑦 dimension – every point moves parallel to the 𝑦-axis when the string vibrates
 𝑦(𝑥, 𝑡) = the position of the string at point 𝑥 at time 𝑡
 The shape of the string is the curve of 𝑦
Boundary value problem
• 𝑦𝑡𝑡 = 𝑎2 𝑦𝑥𝑥 , 0 < 𝑥 < 𝐿 Wave equation
• 𝑦(0, 𝑡) = 𝑦(𝐿, 𝑡) = 0 Endpoints are fixed
• 𝑦(𝑥, 0) = 𝑓(𝑥) Initial position
• 𝑦𝑡 (𝑥, 0) = 𝑔(𝑥) Initial velocity

PROBLEM A PROBLEM B
𝑦𝑡𝑡 = 𝑎2 𝑦𝑥𝑥 , 0 < 𝑥 < 𝐿 𝑦𝑡𝑡 = 𝑎2 𝑦𝑥𝑥 , 0 < 𝑥 < 𝐿
𝑦(0, 𝑡) = 𝑦(𝐿, 𝑡) = 0 𝑦(0, 𝑡) = 𝑦(𝐿, 𝑡) = 0
𝑦(𝑥, 0) = 𝑓(𝑥) 𝑦(𝑥, 0) = 0
𝑦𝑡 (𝑥, 0) = 0 𝑦𝑡 (𝑥, 0) = 𝑔(𝑥)

These equations are found using the separation of variables

 𝑦(𝑥, 𝑡) = 𝑋(𝑥)𝑇(𝑡)

Solutions for problem A Solutions for problem B

Fourier solution for problem A Fourier solution for problem B
∞ ∞
𝒏𝝅𝒙 𝒏𝝅𝒂𝒕 𝒏𝝅𝒙 𝒏𝝅𝒂𝒕
𝒚𝑨 (𝒙, 𝒕) = ∑ 𝒄𝒏 [𝐜𝐨𝐬 ] [𝐬𝐢𝐧 ] 𝒚𝑩 (𝒙, 𝒕) = ∑ 𝒄𝒏 [𝐬𝐢𝐧 ] [𝐬𝐢𝐧 ]
𝑳 𝑳 𝑳 𝑳
𝒏=𝟏 𝒏=𝟏
2 𝐿 𝑛𝜋𝑥 2 𝐿 𝑛𝜋𝑥
with 𝑐𝑛 = ∫ [𝑓(𝑥)] [sin 𝐿 ] 𝑑𝑥
𝐿 0
with 𝑐𝑛 = 𝑛𝜋𝑎 ∫0 [𝑔(𝑥)] [sin 𝐿
] 𝑑𝑥

D’Alembert solution for problem A D’Alembert solution for problem B

𝟏 𝟏
𝒚𝑨 (𝒙, 𝒕) = [𝑭(𝒙 + 𝒂𝒕) + 𝑭(𝒙 − 𝒂𝒕)] 𝒚𝑩 (𝒙, 𝒕) = [𝑯(𝒙 + 𝒂𝒕) + 𝑯(𝒙 − 𝒂𝒕)]
𝟐 𝟐𝒂
𝑥
With F the odd extension of 𝑓 [𝑝 = 2𝐿] With 𝐻(𝑥) = ∫0 𝐺(𝑠). 𝑑𝑠
OR
𝑥+𝑎𝑡
𝟏
𝒚𝑩 (𝒙, 𝒕) = [∫ 𝐺(𝑠). 𝑑𝑠 ]
𝟐𝒂 𝑥−𝑎𝑡
With 𝐺 the odd extension of 𝑔 [𝑝 = 2𝐿]

 Series solution: 𝒚(𝒙, 𝒕) = 𝒚𝑨(𝒙, 𝒕) + 𝒚𝑩 (𝒙, 𝒕)

Method with separation of variables:
1) Set equal to 𝑋(𝑥 )𝑇(𝑡)
2) Substitute into the heat transfer or wave function
𝑋′′
3) Simplify to get the form 𝑋
= (𝑖. 𝑡. 𝑜. 𝑇 ′𝑎𝑛𝑑 𝑇) = −𝜆

4) Find a problem for 𝑋 and 𝑇: 𝑋 ′′ + 𝜆𝑋 = 0 and simplify for 𝑇

- ( )𝑇 ′ + ( )𝑇 = −𝜆( )
5) Find eigenvalues and eigenvectors for 𝑋 ′′ + 𝜆𝑋 = 0
6) Substitute back in