Math Methods Exam 2 Formula Sheet

Complex Numbers Power Series

A real series converges if
𝑒𝑖 𝜃 = cos 𝜃 + 𝑖 sin 𝜃 (1a)
𝑎𝑛+1
(𝑐𝑜𝑠 𝜃 + 𝑖 sin 𝜃 ) 𝑛 = cos(𝑛 𝜃 ) + 𝑖 sin(𝑛 𝜃 ) (1b) 1 > 𝜌 = | 𝑥 | lim (9)
𝑛→∞ 𝑎𝑛
ln 𝑧 = ln 𝑟 + 𝑖 ( 𝜃 + 2 𝜋𝑛) (1c)
And a complex series has a radius of convergence, 𝑅 by

1 𝑎𝑛+1
= lim (10)
Hyperbolic Functions 𝑅 𝑛→∞ 𝑎𝑛

Where points on the boundary must be tested explicitly

−
𝑒𝑖 𝑥 𝑒 −𝑖 𝑥 𝑒𝑖 𝑥 + 𝑒 −𝑖 𝑥
sin 𝑥 = cos 𝑥 = (2a) Taylor Series
2𝑖 2
𝑒 𝑥 − 𝑒 −𝑥 𝑒𝑥 + 𝑒 −𝑥
sinh 𝑥 = cosh 𝑥 = (2b)
2 2 ( 𝑥 − 𝑎) 2 ′′
sinh(− 𝑥 ) = − sinh 𝑥 cosh(− 𝑥 ) = cosh 𝑥 (2c) 𝑓 ( 𝑥 ) = 𝑓 (𝑎) + ( 𝑥 − 𝑎) 𝑓 ′ (𝑎) + 𝑓 (𝑎) + . . . (11a)
2
𝑑 sinh 𝑥 𝑑 cosh 𝑥
= cosh 𝑥 = sinh 𝑥 (2d) (𝑥 −𝑎) 𝑛+1
𝑑𝑥 𝑑𝑥 𝑅𝑛 ( 𝑥 ) = 𝑓 (𝑛+1) ( 𝜉 ) (11b)
(𝑛 + 1)!
sin(𝑖 𝑥) = 𝑖 sinh 𝑥 cos(𝑖 𝑥 ) = cosh 𝑥 (2e)
sinh(𝑖 𝑥) = 𝑖 sin 𝑥 cosh(𝑖 𝑥 ) = cos 𝑥 (2f)
1 = cosh2 − sinh2 (2g) Useful MacLaurin Series

𝑥3 𝑥5
sin 𝑥 = 𝑥 − + − ... (12a)
Sums of Finite Series 3! 5!
𝑥2 𝑥4
Arithmetic series (constant difference 𝑑): cos 𝑥 = 1 − + − ... (12b)
2! 4!
𝑁 𝑥 2
𝑆𝑁 = (2𝑎0 + ( 𝑁1 ) 𝑑) (3) 𝑒𝑥 = 1 + 𝑥 + +... (12c)
2 2!
Geometric Series (constant ratio 𝑟): (𝑛 − 1) 2 (𝑛 − 2) (𝑛 − 1) 3
(1 + 𝑥 ) 𝑛 = 1 + 𝑛𝑥 + 𝑛𝑥 + 𝑛𝑥 + . . . (12d)
2! 3!
1− 𝑟𝑛
𝑆 𝑁 = 𝑎0 (4)
1−𝑟
Types of Matrices
Difference Method (𝑢𝑛 = 𝑓 (𝑛) − 𝑓 (𝑛 − 1) for some function 𝑓 (𝑛)):

𝑆 𝑁 = 𝑓 ( 𝑁 ) − 𝑓 (0) (5) Symmetric 𝐴𝑇 = 𝐴 || Orthogonal 𝐴𝑇 = 𝐴−1 || Hermitian 𝐴 = 𝐴†

Unitary 𝐴† = 𝐴−1 || Normal 𝐴† 𝐴 = 𝐴𝐴† so [ 𝐴, 𝐴† ] = 0.
Useful relations:
𝑁 Similarity Transformation
∑︁ 1
𝑛= 𝑁 ( 𝑁 + 1) (6a)
2
𝑛=1
𝑁
𝐴′ = 𝑆 −1 𝐴𝑆 (13)
∑︁ 1
𝑛2 = 𝑁 ( 𝑁 + 1) (2𝑁 + 1) (6b)
6
𝑛=1
𝑁
∑︁ 1 2 Matrix Decompositions
𝑛3 = 𝑁 ( 𝑁 + 1) 2 (6c)
4 LU Decomposition:
𝑛=1
• Decompose into lower and upper triangular matrices 𝐿 and 𝑈, 𝐿
has 1’s on diagonal
• 𝐴 = 𝐿𝑈
Infinite Series • Use 𝐿 𝑦® = 𝑏® and use backsubstituion to solve for 𝑦® , then solve
𝑈 𝑥® = 𝑦® similarly
Comparison test: if "friend" series converges and is strictly greater, then Cholesky Decomposition:
series converges. Flip reasoning for diverging series. • Works for Hermitian matrices with eigenvalues ≥ 0
D’Alembert ratio test (converges if 𝜌 < 1, diverges if 𝜌 > 1, inconclusive • 𝐴 = 𝐿𝐿†, where 𝐿 is lower-triangular
otherwise) SVD Decomposition:
  • 𝐴 = 𝑈𝑆𝑉 † where
𝑢𝑛+1 • The columns of 𝑈 are the eigenvectors of 𝐴𝐴†
𝜌 = lim (7)
𝑛→∞ 𝑢𝑛 • The columns of 𝑉 are the eigenvectors
√ of 𝐴 𝐴
†

• 𝑆 is a diagonal matrix where 𝑠𝑖 = 𝜆𝑖 where 𝜆𝑖 are the eigenval-

Ratio comparison test: ues of either 𝑈 or 𝑉
• Friend series 𝑣𝑛
Í
• Solution is 𝑥® = 𝑉 𝑆𝑈 ® where the 𝑆¯ is diagonal with each entry
¯ † 𝑏,
• Converges if 𝑢𝑛+1
𝑢 𝑣𝑛+1 Í
𝑛
≤ 𝑣𝑛 for 𝑛 > 𝑁 if 𝑣𝑛 is convergent. being 𝑠1
• Converges if 𝑢𝑛+1
𝑢 𝑣𝑛+1 Í 𝑖
𝑛
≥ 𝑣𝑛 for 𝑛 > 𝑁 if 𝑣𝑛 is divergent.
Integral test, converges if sum∫ is 𝑢𝑛 = 𝑓 (𝑛) where 𝑓 (𝑛) is monotoni- Fourier Series
∞
cally decreasing, and integral 0 𝑓 ( 𝑥 ) 𝑑 𝑥 converges.
Cauchy root test (converges if 𝜌 < 1, diverges if 𝜌 > 1, inconclusive
∞ 
otherwise)
   
𝑎0 ∑︁ 2 𝜋𝑛𝑥 2 𝜋𝑛𝑥
𝑓 (𝑥) = + 𝑎𝑛 cos + 𝑏𝑛 sin (14a)
2 𝐿 𝐿
𝜌 = lim (𝑢𝑛 ) 1/𝑛 (8) 𝑛=1
𝑛→∞ ∫ 𝑥0 +𝐿  
2 2 𝜋𝑛𝑥
𝑎𝑛 = 𝑓 ( 𝑥 ) cos 𝑑𝑥 (14b)
𝐿 𝑥0 𝐿
∫ 𝑥0 +𝐿  
2 2 𝜋𝑛𝑥
Alternating Series 𝑏𝑛 = 𝑓 ( 𝑥 ) sin 𝑑𝑥 (14c)
𝐿 𝑥0 𝐿
Alternating series converges if:
• 𝑢𝑛 → 0 as 𝑛 → ∞ Where the 𝑎0 case must be evaluated separately (plug in 𝑛 = 0 into the
• 𝑢𝑛 < 𝑢𝑛−1 for all 𝑛 > 𝑁 for some finite 𝑁 𝑎𝑛 formula).
 Math Methods Exam 2 Formula Sheet

Complex Fourier Series Linear Recurrence Relations

In which you are given 𝑢𝑛+𝑑 in terms of smaller-𝑛 terms (𝑢𝑛 , 𝑢0 , etc.).
∞ The steps to solve:
2 𝜋𝑖𝑛𝑥
• Create the complementary equation. To do this, sub in 𝑢𝑛 = 𝐴𝜆𝑛
∑︁
𝑓 (𝑥) = 𝑐𝑛 𝑒 𝐿 (15a)
𝑛=−∞ so that 𝑢𝑛+𝑑 → 𝜆𝑑 . Constants unchanged.
∫ 𝑥0 +𝐿 • Solve the complementary equation for 𝜆
1 2 𝜋𝑖𝑛𝑥
• Then, 𝑢𝑛 = 𝐴(𝜆1 ) 𝑛 + 𝐵(𝜆2 ) 𝑛 , generalized for more roots.
𝑐𝑛 = 𝑓 ( 𝑥 )𝑒 − 𝐿 𝑑𝑥 (15b)
𝐿 𝑥0 • Use boundary/initial conditions to determine 𝐴 and 𝐵

Parseval’s Theorem Laplace Transforms

For Fourier Series Parseval’s Theorem is: Definition of Laplace Transform:
∫ 𝑥0 +𝐿 ∞
1 ∑︁ ∫ ∞
| 𝑓 ( 𝑥 ) |2 𝑑 𝑥 = |𝑐𝑛 | 2 (16a) L [ 𝑓 (𝑡 ) ] = 𝑓¯ (𝑠) = 𝑓 (𝑡 )𝑒 −𝑠𝑡 𝑑𝑡 (24)
𝐿 𝑥0 𝑛=−∞ 0
𝑎02 ∞
1 ∑︁
To solve a differential equation, Laplace transform both sides, then solve
= + (𝑎𝑛2 + 𝑏𝑛2 ) (16b)
4 2
𝑛=0 for 𝑓¯ (𝑠). Then, inverse Laplace transform to get 𝑓 (𝑡 ). Useful relation:

For Fourier Transforms Parseval’s Theorem is: L [ 𝑓 ′ (𝑡 ) ] = 𝑠 𝑓¯ (𝑠) − 𝑓 (0) (25a)

∫ ∞ ∫ ∞ L [ 𝑓 ′′ (𝑡 ) ] = 𝑠 2 𝑓¯ (𝑠) − 𝑠 𝑓 (0) − 𝑓 ′ (0) (25b)
| 𝑓 ( 𝑥 ) |2 𝑑 𝑥 = | 𝑓˜ (𝑘 ) | 2 𝑑𝑘 (17)
−∞ −∞ Which can be generalized to higher order derivatives.

Fourier Transforms
Green’s Functions

1
∫ ∞ Step-by-step to solve an ODE with a Green’s Function:
𝑓˜ ( 𝜔) = √ 𝑓 (𝑡 )𝑒 −𝑖 𝜔𝑡 𝑑𝑡 (18a) • Solve homogenous differential equation using complementary
2 𝜋 −∞ equation. n
𝑦 ( 𝑥 ) 𝑥 < 𝑥′
∫ ∞
1 • 𝐺 ( 𝑥, 𝑥 ′ ) = 𝑦𝑐,1 ( 𝑥 ) 𝑥 > 𝑥 ′
𝑓 (𝑡 ) = √ 𝑓˜ ( 𝜔)𝑒𝑖 𝜔𝑡 𝑑𝑡 (18b) 𝑐,2
2 𝜋 −∞ • Where the different complementary solutions differ by having dif-
F [ 𝑓 ′ (𝑡 ) ] = 𝑖 𝜔 𝑓˜ ( 𝜔) (18c) ferent constants.
∫ 𝑡
• Use boundary conditions (of 𝑦 ( 𝑥 )) to determine two constants
1 ˜ (the ones in the interval containing the b.c.’s)
F[ 𝑓 (𝑠) 𝑑𝑠] = 𝑓 ( 𝜔) + 2 𝜋𝑐 𝛿 ( 𝜔) (18d)
𝑖𝜔 • Use continuity to find expression for third constant: 𝐺 + ( 𝑥, 𝑥 ′ ) =
1 𝜔 𝐺 − ( 𝑥, 𝑥 ′ )
F [ 𝑓 (𝑎𝑡 ) ] = 𝑓˜ ( ) (18e) • Use discontinuity to find final constant: 𝐺 + ( 𝑥, 𝑥 ′ ) −𝐺 − ( 𝑥, 𝑥 ′ ) =
𝑎 𝑎 1 (for higher order 𝑛 the discontinuity is in the 𝑛 − 1’th derivative
F [ 𝑓 (𝑡 + 𝑎) ] = 𝑒𝑖𝑎 𝜔 𝑓˜ ( 𝜔) (18f) and has a discontinuity 𝑎𝑛 1( 𝑥 ′ ) ), where 𝑎𝑛 ( 𝑥 ′ ) is the coefficient
F [𝑒 𝑎𝑡 𝑓 (𝑡 ) ] = 𝑓˜ ( 𝜔 + 𝑖 𝛼) (18g) of the highest derivative in the differential equation
Once the Green’s Function is found, the solution is
∫ 𝑏
Convolution and The Convolution Theorem 𝑦(𝑥) = 𝐺 ( 𝑥, 𝑥 ′ ) 𝑓 ( 𝑥 ′ ) 𝑑 𝑥 ′ (26)
𝑎
A convolution is defined as: Where the bounds for instance may go from [0, 𝑡 ].
∫ ∞ ∫ ∞
ℎ(𝑧) = 𝑓 ∗ 𝑔 = 𝑓 ( 𝑥 )𝑔 (𝑧 − 𝑥 ) 𝑑 𝑥 = 𝑓 (𝑧 − 𝑥 )𝑔 ( 𝑥 ) 𝑑 𝑥
−∞ −∞
(19) Hilbert Space and Adjoint of Linear Operator
The convolution theorem states that A Hilbert space is defined as a space of functions satisfying the inner
√ product:
ℎ˜ (𝑘 ) = 2 𝜋 𝑓˜ (𝑘 ) 𝑔˜ (𝑘 ) (20a)
1 ˜
∫ 𝑏
F [ 𝑓 ( 𝑥 )𝑔 ( 𝑥 ) ] = √ 𝑓 (𝑘 ) ∗ 𝑔˜ (𝑘 ) (20b) ⟨ 𝑓 |𝑔⟩ = 𝑓 ∗ ( 𝑥 )𝑔 ( 𝑥 )𝜌( 𝑥 ) 𝑑 𝑥 (27)
2𝜋 𝑎

The adjoint of a linear operator is:

Linear ODEs D E
⟨ 𝑓 | L𝑔⟩ = L † 𝑓 𝑔 + 𝑏.𝑡. (28)
A linear ODE will have the form (for second order, can be generalized to
more)
Where 𝑏.𝑡. are the "boundary terms". If L † = L, then the operator is
′′ ′
𝑓 ( 𝑥 ) = 𝑎2 𝑦 ( 𝑥 ) + 𝑎1 𝑦 ( 𝑥 ) + 𝑎0 𝑦 ( 𝑥 ) (21) self-adjoint. If 𝑏.𝑐. = 0, then the linear operator is Hermitian.
A Green’s function can be expanded in terms of eigenfunctions of a
Hermitian operator:
Where 𝑓 ( 𝑥 ) also includes any constants. This has the complementary
equation ∞
∑︁ 1
2
𝐺 ( 𝑥, 𝑥 ′ ) = 𝑦ˆ 𝑛 ( 𝑥 ) 𝑦ˆ 𝑛∗ (𝑧) (29)
0 = 𝑎2 𝜆 + 𝑎1 𝜆 + 𝑎0 (22) 𝑛=0
𝜆𝑛

Which can be solved for 𝜆 to find the roots. There are three possible
cases:
• Real, distinct roots: 𝑦𝑐 ( 𝑥 ) = 𝑐1 𝑒𝜆1 𝑥 + 𝑐2 𝑒𝜆2 𝑥
• Repeated roots: 𝑦𝑐 ( 𝑥 ) = 𝑐1 𝑒𝜆𝑥 + 𝑐2 𝑥𝑒𝜆𝑥
• Imaginary roots (𝜆1,2 = 𝛼 ± 𝛽𝑖): 𝑦𝑐 ( 𝑥 ) = 𝑒 𝛼𝑥 [𝑐1 cos(𝛽 𝑥 ) + Green’s Functions as a Superposition of a Hermitian Operator
𝑐2 sin(𝛽 𝑥 ) ]
To find the particular solution, plug in 𝑦 = 𝐶 𝑓 ( 𝑥 ) into the RHS (keeping ∞
∑︁ 1
the LHS the same) of the differential equation, then solve for 𝐶. Then, 𝐺 ( 𝑥, 𝑥 ′ ) = 𝑦ˆ 𝑛 ( 𝑥 ) 𝑦ˆ 𝑛∗ ( 𝑥 ′ ) (30)
𝜆𝑛
𝑛=0
𝑦𝑝 ( 𝑥 ) = 𝐶 𝑓 ( 𝑥 ) (23a)
Where 𝜆𝑛 are the eigenvalues satisfying the eigenvalues equation
𝑦 ( 𝑥 ) = 𝑦𝑐 ( 𝑥 ) + 𝑦 𝑝 ( 𝑥 ) (23b)
L 𝑦𝑖 ( 𝑥 ) = 𝜆𝑖 𝑦𝑖 ( 𝑥 ) for the linear operator L corresponding to the dif-
ferential equation and for orthonormal eigenfunctions 𝑦𝑖 ( 𝑥 ).
 Math Methods Exam 2 Formula Sheet

Sturm-Liouville DE’s Famous PDE’s

S-L DE’s correspond to a Hermitian differential operator, and thus have 1 𝜕2 𝑢
real eigenvalues/eigenfunctions. • Wave Equation: ∇ 2 𝑢 =
𝑐 2 𝜕𝑡 2
2
• Diffusion Equation: ∇ 2 𝑢 = 12 𝜕 𝑢2
0 = 𝑝 ( 𝑥 ) 𝑦 ′′ ( 𝑥 ) + 𝑟 ( 𝑥 ) 𝑦 ′ ( 𝑥 ) + 𝑞 ( 𝑥 ) 𝑦 ( 𝑥 ) + 𝜆𝜌𝑦 ( 𝑥 ) (31a) 𝑐 𝜕𝑡
′ ′ • Laplace Equation: ∇2 𝑢 = 0
0 = ( 𝑝𝑦 ) + 𝑞𝑦 + 𝜆𝜌𝑦 (31b)
• Poisson Equation: ∇ 2 𝑢 = 𝜌( 𝑟® )
′
𝑟 (𝑥) = 𝑝 (𝑥) (31c) • Schrödinger Equation: − 2𝑚ℏ2 2
∇ 𝑢 + 𝑉 ( 𝑟® )𝑢 = 𝑖ℏ 𝜕𝑢
𝜕𝑡
Any second order DE of the form 𝑝𝑦 ′′ + 𝑟 𝑦 ′ + 𝑞𝑦 + 𝜆𝜌𝑦 = 0 can be
converted to a S-L DE by multiplying it by:

𝑟 (𝑢) − 𝑝 ′ (𝑢)
∫ 𝑥 
𝐹 ( 𝑥 ) = exp 𝑑𝑢 (31d) General Solutions of PDE’s
𝑝 (𝑢)
𝜕𝑢
For a first-order PDE of two variables of the form 𝐴( 𝑥, 𝑦) 𝜕𝑥 +
𝜕𝑢
𝐵( 𝑥, 𝑦) 𝜕𝑦 = 0, integrate
Series solutions of DE’s
For finding the solution of a DE around an ordinary point, plug in: 𝑑𝑥 𝑑𝑦
= (37a)
𝐴 𝐵
∞
∑︁
𝑦 (𝑧) = 𝑎𝑛 𝑧 𝑛 (32) Which will give an equation of 𝑥, 𝑦, and the integration constant, which
𝑛=0 we set to 𝑝 and solve for. Then, find 𝑝 along the boundary, solve for
the remaining variable, then plug it into the boundary condition. Then,
Then, reorganize the series such that all the series start at the same initial substitute in the general form for 𝑝.
value, then combine and solve for the recurrence relation. Then, write For a second-order PDE of two variables of the form
2 𝜕2 𝑢 2
out the first few terms, collect terms and use them to define the series. 𝐴( 𝑥, 𝑦) 𝜕 𝑢2 + 𝐵( 𝑥, 𝑦) 𝜕𝑥𝜕𝑦 + 𝐶 ( 𝑥, 𝑦) 𝜕 𝑢2 = 0, solutions are of the
However, if the point is a regular singular point, use the Frobenius 𝜕𝑥 𝜕𝑦
method, using this form: form
∞ 𝑢( 𝑥, 𝑦) = 𝑓 ( 𝑥 + 𝜆+ 𝑦) + 𝑔 ( 𝑥 + 𝜆 − 𝑦) (37b)
∑︁ √
𝑦 (𝑧) = 𝑧 𝜎 𝑎𝑛 𝑧 𝑛 (33) −𝐵 ± 𝐵2 − 4𝐴𝐶
𝑛=0 𝜆± = (37c)
2𝐴
After plugging this in, taking 𝑛 = 0 for each series will result in the If the expression under the square root is positive, the PDE is hyperbolic.
indicial equation, which when solved will give you a relation which must If it’s negative, the PDE is elliptic, and if its 0 the PDE is parabolic.
be satisfied. Then, follow the same steps as above to determine the series
solution to the DE.

Solutions to Special Function DE’s

Legendre DE: Soln’s are Legendre Polynomials, orthogonal on [ −1, 1] Characteristic Curves of PDE’s
The characteristic curves of a first-order PDE are the set of points where
1 𝑑ℓ
𝑃ℓ ( 𝑥 ) = ( 𝑥 2 − 1) ℓ (34a) 𝑑𝑦 𝐵( 𝑥, 𝑦)
2 𝑙! 𝑑 𝑥 ℓ
ℓ
= (38a)
∞ 
1 ∑︁ 𝑟 ′ ℓ 𝑑𝑥 𝐴( 𝑥, 𝑦)

1
= 𝑃ℓ (cos 𝜃 ) (34b)
| 𝑟® − 𝑟®′ | 𝑟 𝑟 is constant, equivalently where 𝑝 is constant.
𝑙=0

Associated Legendre DE: Soln’s are Associated Legendre Polynomials

(−1) 𝑚   𝑚/2 𝑑 ℓ+𝑚  ℓ

𝑃ℓ𝑚 ( 𝑥 ) = 1 − 𝑥2 𝑥2 − 1 (34c)
2ℓ ℓ! 𝑑 𝑥 ℓ+𝑚 Dirichlet, Neumann, and Cauchy Boundary Conditions
Spherical Harmonics: A second-order PDE can have three types of boundary conditions:
• Dirichlet: 𝑢 is specified on the boundary
 1/2
• Neumann: 𝜕𝑛
 𝜕𝑢
2𝑙 + 1 (𝑙 − 𝑚)! = ∇𝑢 · 𝑛® is specified on the boundary, where 𝑛ˆ is
𝑌𝑙𝑚 ( 𝜃 , 𝜙) = (−1) 𝑚 𝑃ℓ𝑚 (cos 𝜃 )𝑒𝑖𝑚𝜙 (34d) normal to the boundary
4 𝜋 (𝑙 + 𝑚)!
• Cauchy: Both 𝑢 and 𝜕𝑛 𝜕𝑢
are specified on the boundary
Bessel DE: Soln’s are Bessel Function of the First and Second Kind: For a solution to be unique for a PDE, it must be specified by the type of
∞
B.C. corresponding to its type:
(−1) 𝑛 • Hyperbolic: open boundary, Cauchy conditions.
∑︁  𝑥  𝜈+2𝑛
𝐽𝜈 ( 𝑥 ) = (34e) • Parabolic: open boundary, Dirichlet or Neumann conditions.
𝑛!Γ (𝜈 + 𝑛 + 1) 2
𝑛=0 • Elliptic: closed boundary, Dirichlet or Neumann conditions
𝐽𝜈 ( 𝑥 ) cos(𝜈 𝜋 ) − 𝐽−𝜈 ( 𝑥 )
𝑌𝜈 ( 𝑥 ) = (34f)
sin(𝜈 𝜋 )
𝐽−𝜈 ( 𝑥 ) = (−1) 𝜈 𝐽𝜈 ( 𝑥 ) (34g)

Where for integer 𝜈 the solution is a linear combination of 𝐽𝜈 and 𝑌𝜈 , Separation of Variables
and for non-integer 𝜈 the solution is a linear combination of 𝐽𝜈 and 𝐽−𝜈 .
Assume form is separable, i.e. can be written as

Gamma Function; Stirling’s Approximation 𝑢( 𝑥, 𝑦) = 𝑋 ( 𝑥 )𝑌 ( 𝑦) (40a)

Gamma Function: or, 𝑢( 𝑥, 𝑦, 𝑧) = 𝑋 ( 𝑥 )𝑌 ( 𝑦) 𝑍 (𝑧) for 3D Cartesian; 𝑢(𝑟 , 𝜃 , 𝜙) =

𝑅 (𝑟 )Θ( 𝜃 )Φ( 𝜙) for 3D spherical, etc. Plug in this solution, then divide
∫ ∞ by the solution to get three terms which are each only dependent on one
Γ (𝑛) = 𝑥 𝑛−1 𝑒 −𝑥 𝑑 𝑥 (35a) coordinate. Bring over one term, then set it and any other remaining
0 terms equal to a constant; repeat if there are multiple terms on one side.
𝑛! = Γ(𝑛 + 1) (35b) Solve the reulting ODEs, and apply B.C.’s.
Γ(𝑧 + 1) = 𝑧Γ(𝑧) (35c)

Stirling’s Approximation(s):
√
𝑛! ∼ 2 𝜋𝑛𝑛𝑛 𝑒 −𝑛 (35d) Integral transform methods for PDE’s (Laplace and Fourier)
ln(𝑛!) ≈ 𝑛 ln(𝑛) − 𝑛 (35e)
Use Fourier or Laplace transform to transform PDE such that in that
space it is easier to solve. Solve, then transform back.
 Math Methods Exam 2 Formula Sheet

Green’s Function Solutions for PDE’s (Method of Images) Laurent Series of Complex Functions

When evaluating a PDE within a volume/area, can sometimes use method If 𝑓 (𝑧) has a pole of order 𝑝 at 𝑧 = 𝑧0 , but is otherwise analytic
of images to solve the problem: within the circle, we use the Laurent series, which is an expansion of
• For each 𝛿-function source inside the boundary, add a source 𝑔 (𝑧) = (𝑧 − 𝑧0 ) 𝑝 𝑓 (𝑧):
reflected over the boundary with opposite charge
• For a Laplacian, the fundamental solution is − 4 𝜋 |𝑟®1−𝑟® | 𝑓 (𝑧) =
𝑎− 𝑝
+...+
𝑎 −1
+ 𝑎0 + 𝑎1 (𝑧 − 𝑧0 ) + . . . (47a)
0
• The Green’s function is the sum of the fundamental solution and (𝑧 − 𝑧0 ) 𝑝 𝑧 − 𝑧0
the associated solutions to each charge
• The solution is then:
∫ ∫ Residue Theorem
𝜕𝐺
𝑢( 𝑟®0 ) = 𝐺 ( 𝑟®, 𝑟®0 )𝜌( 𝑟® ) 𝑑𝑉 𝑟® + 𝑓 ( 𝑟® ) 𝑑𝑆 𝑟® (41a) The residue is the 𝑎 −1 coefficient in the Laurent expansion. The Residue
𝑉 𝑆 𝜕𝑛 Theorem states:
∮ ∑︁
𝑓 (𝑧) 𝑑𝑧 = 2 𝜋𝑖 𝑅𝑗 (48a)
Radius of Convergence of Complex Functions 𝐶 𝑗

A power series for 𝑓 (𝑧) can be written as Where 𝑅 𝑗 is the residue for each pole inside 𝐶. To find the residue
either:
∞
∑︁ • Define 𝑧 = 𝑧0 + 𝜉 for 𝑓 (𝑧), then binomally expand for small 𝜉 ,
𝑓 (𝑧) = 𝑎𝑛 𝑧 𝑛 (42a) 𝑎 −1 is the coefficient of 1/ 𝜉
𝑛=0 • For a simple pole, 𝑅 (𝑧0 ) = lim 𝑧→𝑧0 [ (𝑧 − 𝑧0 ) 𝑓 (𝑧) ]
1 𝑎𝑛+1 • For a pole of order𝑚:
= lim = lim | 𝑎𝑛 | 1/𝑛 (42b) 1 𝑑 (𝑚−1)

𝑅 𝑛→∞ 𝑎𝑛 𝑛→∞ 𝑅 (𝑧0 ) = lim 𝑧→𝑧0 (𝑚−1) ! 𝑑𝑧 (𝑚−1) [ (𝑧 − 𝑧0 ) 𝑓 (𝑧) ]
𝑚
∫2
Where 𝑅 is the radius of convergence. For integrals of the form 0 𝜋 𝑓 (cos 𝜃 , sin 𝜃 ) 𝑑 𝜃, write them in complex
form, and use 𝑧 = 𝑒𝑖 𝜃 , then evaluate the complex integral over the unit
Differentiability of Complex Functions circle using residue form. ∫∞
For infinite integrals of the form −∞ 𝑓 ( 𝑥 ) 𝑑 𝑥, evaluate on a semicircular
A complex function 𝑓 (𝑧) = 𝑢( 𝑥, 𝑦) + 𝑖𝑣 ( 𝑥, 𝑦) is differentiable if it contour in the upper half-plane of radius 𝑅, then make the semicircular
satisfies the Cauchy-Riemann relations: arc vanish by taking 𝑅 → ∞. The conditions for this are: finite amount
of poles in the region, lim |𝑧|→∞ [ 𝑧 𝑓 (𝑧) ] = 0, integrals over (−∞, 0]
𝜕𝑢
=
𝜕𝑣
(43a) and [0, ∞) exist.
𝜕𝑥 𝜕𝑦
𝜕𝑣 𝜕𝑢 Binomial Expansion
=− (43b)
𝜕𝑥 𝜕𝑦

A function is analytic at a point if it is differentiable and single valued at 𝑛  

∑︁ 𝑛
that point. An analytic function’s real and imaginary parts satisfy: ( 𝑥 + 𝑦) 𝑛 = 𝑥 𝑛−𝑘 𝑦 𝑘 (49a)
𝑘
𝑘=0
∇2 𝑢 = ∇2 𝑣 = 0 (43c)
 
𝑛 𝑛!
= (49b)
∇𝑢 · ∇𝑣 = 0 (43d) 𝑘 𝑘!(𝑛 − 𝑘 )!
| ∇𝑢| = | ∇𝑣 | (43e)

Cartesian Tensors
Zeroes, Limits, Poles and Branch Points of Complex Functions Cartesian tensors satisfy the conditions that:
• They are orthogonal, 𝐿𝑖𝑘 𝐿 𝑗 𝑘 = 𝛿𝑖 𝑗
• Zero: Solutions to 𝑓 (𝑧) = 0. If 𝑓 (𝑧) has the form • Under a rotation of axes, they transform as: 𝑥𝑖′ = 𝐿𝑖 𝑗 𝑥 𝑗
𝑓 (𝑧) = (𝑧 − 𝑧0 ) 𝑛 𝑔 (𝑧), the zero is of order 𝑛. Rotation about the 𝑧 axis is given by:
• Limit: Take limits as normal, but for 𝑙𝑖𝑚𝑧→∞ use substitution
𝜉 = 𝑧1 , then take lim 𝜉 𝑡𝑜0
!
cos 𝜃 sin 𝜃 0
• Pole: Points where 𝑓 (𝑧) is not analytic. If 𝑓 (𝑧) has the form 𝐿 = − sin 𝜃 cos 𝜃 0 (50a)
𝑔 (𝑧) 0 0 1
𝑓 (𝑧) = (𝑧−𝑧 ) 𝑛 , the pole is of order 𝑛. If there is no finite 𝑛
0
(such as in functions defined as infinite series, like 𝑒 𝑧 ) the pole is The outer product of a tensor is:
an essential singularity.
• Branch Point: A singularity where the function is multivalued. 𝑇𝑖 𝑗 = 𝑢® ⊗ 𝑣® = 𝑢𝑖 𝑣 𝑗 (51)
Can be proved by evaluating 𝑓 (𝑒𝑖 𝜃 ) for 𝜃 = 0, 2 𝜋 where you will
get different answers.

Useful Tensor Identities

Cauchy’s Theorem, Integral Formula, and Inequality
The gradient of a tensor (which is also a tensor), divergence and curl is:
• Cauchy’s Theorem: If 𝑓 (𝑧)∮is analytic on every point within
and on a boundary curve, then 𝐶 𝑓 (𝑧) 𝑑𝑧 = 0. 𝜕𝑣𝑖
∇𝑣𝑖 = (52a)
• Cauchy’s Theorem: If 𝑓 (𝑧) is analytic on every point within 𝜕𝑥 𝑗
and on a boundary curve, then for a point 𝑧0 within the boundary,
∮ 𝑓 (𝑧) 𝜕𝐹 𝑗
𝑓 (𝑧0 ) = 2 1𝜋𝑖 𝐶 𝑧−𝑧 𝑑𝑧, and the derivative ∇·𝐹= (52b)
∮ 0 𝑓 (𝑧) 𝜕𝑥 𝑗
𝑓 (𝑛) (𝑧0 ) = 2𝑁𝜋𝑖! 𝐶 𝑛+1 𝑑𝑧 𝜕𝐹𝑘
(𝑧−𝑧0 )
∇ × 𝐹 = 𝜖𝑖 𝑗 𝑘 (52c)
• Cauchy’s Inequality: If 𝑓 (𝑧) is analytic inside and on a circle 𝜕𝑥 𝑗
of radius 𝑅 centered on 𝑧0 , then for some constant 𝑀 > | 𝑓 (𝑧) |,
| 𝑓 (𝑛) (𝑧0 ) ≤ 𝑀𝑛!
𝑅𝑛 Generally:

Taylor Series of Complex Functions 𝑎® · 𝑏® = 𝛿𝑖 𝑗 𝑎𝑖 𝑏 𝑗 (52d)

𝑎𝑖 = 𝜖𝑖 𝑗 𝑘 𝑏 𝑗 𝑐𝑘 = 𝑏® × 𝑐® (52e)
If f (z) is analytic inside and on a circle centered on 𝑧 = 𝑧0, and 𝑧 is a
point inside the circle, then 𝜖𝑖 𝑗 𝑘 𝜖 𝑘𝑙𝑚 = 𝛿𝑖𝑙 𝛿 𝑗𝑚 − 𝛿𝑖𝑚 𝛿 𝑗𝑙 (52f)
∞
∑︁
𝑛
𝑓 (𝑧) = 𝑎𝑛 (𝑧 − 𝑧0 ) (46a)
𝑛=0
𝑓 (𝑛) (𝑧0 )
𝑎𝑛 = (46b)
𝑛!