Standard form of Linear First-order ODE: Standard form of Linear Second Order ODEs: Matrix Operations:

𝑦 ′ + 𝑝(𝑥)𝑦 = 𝑟(𝑥) 𝑦 ′′ + 𝑝(𝑥)𝑦 ′ + 𝑞(𝑥)𝑦 = 𝑟(𝑥) Addition, [𝐴] + [𝐵] = [𝐶], requires same size:
 Homogenous in the case, 𝐫(𝐱) = 𝟎: Homogenous in the case, 𝒓(𝒙) = 𝟎 𝐶𝑖𝑗 = 𝐴𝑖𝑗 + 𝐵𝑖𝑗
Separating Variables and Integrating: General solution: Multiplication, [𝐴][𝐵] = [𝐶], requires inner dimension to agree
𝑦(𝑥) = 𝑐1 𝑦1 + 𝑐2 𝑦2 𝑛
𝑦(𝑥) = 𝑐𝑒 −∫ 𝑝(𝑥)𝑑𝑥
 Non-homogenous, 𝒓(𝒙) ≠ 𝟎, use integrating factor: Non-homogenous in the case, 𝒓(𝒙) ≠ 𝟎 𝐶𝑖𝑗 = ∑ 𝑎𝑖𝑘 𝑏𝑘𝑗
𝐹𝑦 ′ + 𝐹𝑝(𝑥)𝑦 = 𝐹𝑟(𝑥) General solution: 𝑘=1

𝑦(𝑥) = 𝑦ℎ𝑜𝑚𝑜𝑔𝑒𝑛𝑜𝑢𝑠 + 𝑦𝑝𝑎𝑟𝑡𝑖𝑐𝑢𝑙𝑎𝑟 Determinants, requires a square matrix, gives a scalar:

𝐹 ′ = 𝑝(𝑥)𝐹 𝑛
Separating variables and solving for 𝐹:  Linear homo ODE with constant coeffs: det(𝑎𝑖𝑗 ) = ∑(−1)𝑖+𝑘 𝑎𝑖𝑘 𝑀𝑖𝑘
𝐹 = 𝑒 ∫ 𝑝(𝑥)𝑑𝑥 𝑦 ′′ + 𝑎𝑦 ′ + 𝑏𝑦 = 0 𝑘=1
(𝐹𝑦)′ = 𝐹𝑟(𝑥) 𝜆2 + 𝑎𝜆 + 𝑏 = 0 Where: 𝑎𝑖𝑘 is the row and expansion point and 𝑀𝑖𝑘 is the
(𝐹𝑦) = ∫ 𝐹𝑟(𝑥)𝑑𝑥 + 𝑐 −𝑎 ± √𝑎2 − 4𝑏 determinant of the submatrix excluding row I and column k
𝜆=
𝑦(𝑥) = 𝑒 −∫ 𝑝(𝑥)𝑑𝑥
(∫ 𝑟 ∙ 𝑒 ∫ 𝑝(𝑥)𝑑𝑥 𝑑𝑥) + 𝑐𝑒 −∫ 𝑝(𝑥)𝑑𝑥 2 Cramer’s Rule, solving: [𝐴]{𝑥} = {𝑏}
Case 1, Two real distinct roots, 𝜆1 & 𝜆2: 𝑏 𝑎12 𝑎 𝑏1
 Transformation to a separable problem: | 1 | | 11 |
𝑦 𝑦 𝑦 = 𝑐1 𝑒 𝜆1 𝑥 + 𝑐2 𝑒 𝜆2𝑥 𝑏2 𝑎22 𝑎21 𝑏2
Let 𝑦 ′ = 𝑓 ( ) and 𝑢 = 1 𝑥1 = , 𝑥2 =
𝑥 𝑥 Case 2, Repeated real roots, 𝜆1 = 𝜆2 = 𝜆 = − 𝑎 det 𝐴 det 𝐴
𝑦 2
Then 𝑦 ′ = 𝑢′ 𝑥
+ 𝑢 = 𝑓 ( ) = 𝑓(𝑢) 𝑎𝑥 Inverse of a Matrix, Augment matrix with I and row operations:
𝑥 𝑦 = (𝑐1 + 𝑐2 𝑥)𝑒 − 2 [𝐴||𝐼] → [𝐼||𝐴−1 ]
𝑑𝑢 𝑑𝑥 𝑎
∫ =∫ Case 3, Complex roots, 𝜆1,2 = − ± 𝑖𝜔: Inverse of a 2x2:
𝑓(𝑢) − 𝑢 𝑥 𝑎𝑥
2
𝑎22 −𝑎12
 Exact Differential Solution 𝑦 = 𝑒 − 2 (𝐴 cos 𝜔𝑥 + 𝐵 sin 𝜔𝑥) [−𝑎
21 𝑎11 ]
−1
[𝐴] =
Problem of the form:  Euler-Cauchy equations: |[𝐴]|
𝑀(𝑥, 𝑦) + 𝑁(𝑥, 𝑦)𝑦 ′ = 0 𝑥 2 𝑦 ′′ + 𝑎𝑥𝑦 ′ + 𝑏𝑦 = 0 Eigenvalue problem:
𝑀(𝑥, 𝑦)𝑑𝑥 + 𝑁(𝑥, 𝑦)𝑑𝑦 = 0 Assume 𝑦 = 𝑥 𝑚 [𝐴]{𝑥} = 𝜆{𝑥}
Compare with the total differential of 𝑢(𝑥, 𝑦) 𝑚2 + (𝑎 − 1)𝑚 + 𝑏 = 0 ([𝐴] − 𝜆[𝐼]){𝑥} = 0
𝜕𝑢 𝜕𝑢 Case 1, Two real distinct roots: det([𝐴] − 𝜆[𝐼]) = 0
𝑑𝑢 = 𝑑𝑥 + 𝑑𝑦 = 0
𝜕𝑥 𝜕𝑦 𝑦 = 𝑐1 𝑥 𝑚1 + 𝑐2 𝑥 𝑚2 Eigenvalues correspond to the roots of det(A-𝜆I)
Then: Case 2, Real repeated root: Eigenvectors:
𝜕𝑢 𝜕𝑢 𝑦 = (𝑐1 + 𝑐2 ln 𝑥)𝑥 𝑚 Determined by arbitrarily choosing 𝑥1 and determining 𝑥2
𝑀(𝑥, 𝑦) = , 𝑁(𝑥, 𝑦) =
𝜕𝑥 𝜕𝑦 Case 3, Complex Roots: System of ODEs, general form:
Check that: 𝑦 = 𝑐1 𝑥 −𝑏+𝑖𝜔 + 𝑐2 𝑥 −𝑏−𝑖𝜔 {𝑦 ′ } = [𝐴]{𝑦}
𝜕𝑀 𝜕2𝑢 𝜕𝑁 𝜕2𝑢  Non-homogenous Linear ODEs: Find eigenvalues and eigenvectors of a, where 𝜆1 is the
= = =
𝜕𝑦 𝜕𝑦𝜕𝑥 𝜕𝑥 𝜕𝑥𝜕𝑦 Method 1 – Undetermined Coeffs first eigenvalue and {𝑥}(1) is the corresponding eigenvector
Then by integration choose one compare with d/dn: (a) Look for r(x) in table 2.1 choose 𝑦𝑝 and determine the coeffs {𝑦} = 𝑐1 {𝑥}(1) 𝑒 𝜆1𝑡 + 𝑐2 {𝑥}(2) 𝑒 𝜆2𝑡 + ⋯
𝑢(𝑥, 𝑦) = ∫ 𝑀𝑑𝑥 + 𝑘(𝑦) (b) If 𝑦𝑝 is part of 𝑦ℎ multiply by 𝑥 or 𝑥 2 Repeated eigenvalues 𝜆1 = 𝜆2 :
{
𝑢(𝑥, 𝑦) = ∫ 𝑁𝑑𝑦 + 𝑙(𝑥) (c) If 𝑟(𝑥) is a sum of functions in 2.1 choose 𝑦𝑝 to also be a sum of {𝑦} = 𝑐1 {𝑥}(1) 𝑒 𝜆1𝑡 + 𝑐2 ({𝑥}(1) 𝑡 + {𝑢})𝑒 𝜆1𝑡
If the problem is not exact: corresponding fs Where {u} solves the equation:([𝐴] − 𝜆1 [𝐼]){𝑢} = {𝑥}(1)
𝜕𝑀 𝜕𝑁 Method 2 – Variations of Parameters Can also be solved as a higher order ODE by differential operators.
≠
𝜕𝑦 𝜕𝑥 (a) Solve for 𝑦ℎ𝑜𝑚𝑜𝑔𝑒𝑛𝑜𝑢𝑠 = 𝑐1 𝑦1 + 𝑐2 𝑦2 Series solution of ODE with Non-Constant coefficients
Then we use an integrating factor: (b) Determine the wronksian: Assume a solution in the form of a power series and substitute:
𝐹𝑃(𝑥, 𝑦)𝑑𝑥 + 𝐹𝑄(𝑥, 𝑦)𝑑𝑦 = 0 a. 𝑊 = 𝑦1 𝑦2′ − 𝑦2 𝑦1′ ∞ ∞
And force this to be exact so that: (c) Determine 𝑦𝑝 𝑦 = ∑ 𝑎𝑛 𝑥𝑛 , 𝑦′ = ∑(𝑛 + 1)𝑎𝑛+1 𝑥 𝑛
𝜕(𝐹𝑃) 𝜕(𝐹𝑄) 𝑦2 𝑟(𝑥) 𝑦1 𝑟(𝑥) 𝑛=0 𝑛=0
= 𝑦𝑝 = −𝑦1 ∫ 𝑑𝑥 + 𝑦2 ∫ 𝑑𝑥 ∞
𝜕𝑦 𝜕𝑥 𝑊 𝑊
Assume that F is a function of only one variable 𝑦 ′′ = ∑(𝑛 + 2)(𝑛 + 1)𝑎𝑛+2 𝑥 𝑛
𝑛=0
Solve the resulting sum = 0 in terms of n and a
Finally relate 𝑎𝑛 to 𝑎0 or 𝑎0 and 𝑎1 for 2nd order ODE