Chapter 1 Complex Functions and Analytic Functions
Complex Numbers
An ordered pair of real numbers is a complex number.
z = x + iy = \rho{e^{i\phi}}
\phi
is called the argument of z
, denoted as
\phi = \arg(z)
The principal value of the argument \text{Arg}(z)
-\pi < \text{Arg}(z) \leq \pi
Regions
Neighborhood of a Point
|z - z_0| \leq \varepsilon_0
Interior Point
If z_0
and its neighborhood both belong to E
, then z_0
is an interior point of E
.
Boundary Point
Domain
Closure
Simply Connected Domain
Multiply Connected Domain
Complex Functions
Derivative
- Differentiability implies continuity, but continuity does not imply differentiability.
- If the real and imaginary parts are continuous, then the complex function is continuous.
- If the real and imaginary parts are differentiable, the complex function is not necessarily differentiable.
Sufficient Conditions for Derivability
Cauchy-Riemann Equations
If w=f(z)=u(x,y) + iv(x,y)
is differentiable at a point z = x+iy
within a region D
, then at (x,y)
there exists and satisfies the Cauchy-Riemann equations
\frac{\partial{u}}{\partial{x}} = \frac{\partial{v}}{\partial{y}},
\frac{\partial{v}}{\partial{x}} = -\frac{\partial{u}}{\partial{y}}
In polar coordinates, it is
\frac{\partial{u}}{\partial{\rho}} = \frac{1}{\rho}\frac{\partial{v}}{\partial{\phi}},
\frac{\partial{v}}{\partial{\rho}} = -\frac{1}{\rho}\frac{\partial{u}}{\partial{\phi}}
Analytic Functions
Definition: If a function f(z)
is differentiable at every point in the neighborhood of z_0
, then the function f(z)
is said to be analytic at z_0
. If f(z)
is not analytic at z_0
, then z_0
is called a singularity of the function f(z)
.
- Analyticity implies differentiability. Being differentiable at a point and being analytic are not the same.
- If the region
D
is open, then every point and its neighborhood belong to the regionD
, and differentiability and analyticity are equivalent within this region.
Properties of Analytic Functions
Orthogonality
If f(z) = u(x,y) + iv(x,y)
is analytic in the region D
, then the family of curves u(x,y)=C, v(x,y)=D
is orthogonal in D
.
That is
\frac{\partial{u}}{\partial{y}}\frac{\partial{v}}{\partial{y}} + \frac{\partial{u}}{\partial{x}}\frac{\partial{v}}{\partial{x}} = 0
Harmonic Functions
If f(z) = u(x,y) + iv(x,y)
is analytic in the region D
, then u
, v
are harmonic functions in the region D
.
The real and imaginary parts are connected by the Cauchy-Riemann equations
Only the real or imaginary part of the analytic function is needed to determine the analytic function.
Conformal Mapping
Elementary Analytic Functions
Elementary Analytic Functions f(z) |
Analytic Region | Derivative | Period | Single-Valued/Multi-Valued |
---|---|---|---|---|
z^n (n > 0 integer) |
Entire plane | nz^{n-1} |
Non-periodic function | Single-Valued |
z^n (n < 0 integer) |
Entire plane except z = 0 |
nz^{n-1} |
Non-periodic function | Single-Valued |
e^z |
Entire plane | e^z |
2\pi i |
Single-Valued |
\sin z |
Entire plane | \cos z |
2\pi |
Single-Valued |
\cos z |
Entire plane | -\sin z |
2\pi |
Single-Valued |
\sinh z |
Entire plane | \cosh z |
None | Single-Valued |
\cosh z |
Entire plane | \sinh z |
None | Single-Valued |
\ln z |
Single-valued branch | \frac{1}{z} |
Non-periodic function | Multi-Valued |
z^s |
Single-valued branch | z^z(\ln z + 1) |
Non-periodic function | Multi-Valued |
Chapter 2 Integrals of Analytic Functions
Integrals of Complex Functions
Closed Loop Integrals
\int_L f(z) dz = \int_L [u(x, y) dx - v(x, y) dy] + i\int_L [v(x, y) dx + u(x ,y) dy]
Properties
- The modulus of the integral of a complex function is not greater than the integral of the modulus of the integrand
| \int_L f(z) dz | \leq \int_L |f(z)| |dz| \leq \int_L |f(z)| dl
- Estimation of Integrals
If|f(z)| \leq M
, then|\int_L f(z) dz| \leq Ml
Cauchy's Theorem
Cauchy's Theorem for Simply Connected Regions
If the function f(z)
is analytic within and on the boundary line L
of a simply connected region D
, then the integral of f(z)
along L
or any closed curve l
within D
is 0
\oint_L f(z) dz = 0, \oint_l f(z) dz = 0
The integral of an analytic function is path-independent.
Cauchy's Theorem for Multiply Connected Regions
Let L
be the outer boundary of a closed multiply connected region D
, and C_1, C_2, \cdots, C_n
be the inner boundaries, then
\oint_L f(z) dz + \sum_{k=1}^{n} \oint_{C_k}f(z)dz = 0
Theorem of Contour Deformation
Let C_1
and C_2
be two positively oriented simple closed curves, and C_1
is enclosed within C_2
. If the function f
is analytic in the multiply connected region enclosed by C_1
and C_2
, then
\oint_{C_1} f(z) dz = \oint_{C_2} f(z) dz
An Important Integral
\oint_{C_R}\frac{dz}{(z-z_0)^n} = \begin{cases}2\pi i, n = 1\\0, n \neq 1 \end{cases}
C_R
is a positively oriented circle with center z_0
and radius R
.
Cauchy's Integral Formula for Simply Connected Regions
f(z_0) = \frac{1}{2\pi j}\oint_l \frac{f(z)}{ z - z_0} dz
where l
is any positively oriented curve within D
, and z_0
is any point within l
.
Cauchy's Integral Formula for Multiply Connected Regions
f(z_0) = \frac{1}{2\pi j}\oint_{L + C_1 + C_2 + \cdots + C_n} \frac{f(z)}{z-z_0}dz
where L
is the outer boundary, C_1, C_2, \cdots, C_n
are the inner boundaries, and z_0
is any point within D
.
Cauchy's Integral Formula for Unbounded Regions
If f(z)
is analytic outside a closed curve L
and f(z) \rightarrow 0
as |z| \rightarrow \infty
, and z_0
is any point outside L
, then
f(z_0) = \frac{1}{2\pi j}\oint_L \frac{f(z)}{z - z_0} dz
Some Corollaries
Derivatives
If f(z)
is analytic within a simply connected region D
, then its derivatives are also analytic within that region, and
f^{(n)}(z) = \frac{n!}{2\pi j}\oint_C \frac{f\xi}{(\xi - z)^{n+1}} dz ~~~~~~~n = 1, 2, 3, \cdots
Chapter 3 Series of Complex Functions
Properties of Complex Series
Theorem 1
The necessary and sufficient condition for the convergence of \sum_{k=1}^{\infty} w_k
is: \forall \varepsilon > 0, \exists N \in \mathcal{N}, \text{such that} n > N \text{when}, |\sum_{k=n+1}^{n+p}w_k| < \varepsilon
,
Theorem 2
Let w_k = u_k + iv_k
, then the necessary and sufficient condition for the convergence of \sum_{k=1}^{\infty} w_k
is that the real part \sum_{k=1}^{\infty} u_k
and the imaginary part \sum_{k=1}^{\infty} v_k
both converge.
Theorem 2'
Let w_k = u_k + iv_k, S = a + ib
, then the necessary and sufficient condition for the convergence of \sum_{k=1}^{\infty} w_k
to S
is that the real part \sum_{k=1}^{\infty} u_k
converges to a
and the imaginary part \sum_{k=1}^{\infty} v_k
converges to b
.
Theorem 3
The necessary condition for the convergence of the series \sum_{k=1}^{\infty} w_k
is \lim_{k \rightarrow \infty} w_k = 0
Theorem 4
If the series \sum_{k=1}^{\infty} |w_k|
converges, then the series \sum_{k=1}^{\infty} w_k
converges.
Theorem 5
The necessary and sufficient condition for the absolute convergence of \sum_{k=1}^{\infty} w_k
is that \sum_{k=1}^{\infty} u_k
and \sum_{k=1}^{\infty} v_k
both converge absolutely.
Series of Complex Functions
Theorem 6
The necessary and sufficient condition for the convergence of the series of complex functions \sum_{n=0}^N f_n(z)
is that for any point z
in D
, given any \varepsilon > 0
, there exists N(z)
such that when n > N(z)
, for any positive integer p
, we have
|\sum_{k=n+1}^{n+p} f_k(z)| < \varepsilon
Uniform Convergence
\forall \varepsilon > 0
, there exists an N
independent of z
, such that for all z
on the region D
or curve L
: when n > N
,
|\sum_{k=n+1}^{n+p} f_k(z)| < \varepsilon
Then the series \sum_{n=0}^N f_n(z)
is said to converge uniformly on D
or L
.
Majorant Test
If there exists |f_n(z)| \leq M_n
, and the series \sum_{n=1}^{\infty} M_n
converges, then the series \sum_{n=1}^{\infty} f_n(z)
converges absolutely and uniformly.
Power Series
Radius of Convergence
- Ratio Test
R = \lim_{n\rightarrow\infty}|\frac{a_n}{a_{n+1}}|
- Root Test
R = \lim_{n\rightarrow\infty}\frac{1}{|a_n|^{\frac{1}{n}}}
Properties
Property 1
Power series converge absolutely and uniformly within the circle of convergence.
Property 2
The sum, difference, and product of convergent series form a new series that converges within |z| < min(R_1, R_2)
.
Property 3
Let the radius of convergence of the power series be R
, then its sum function is analytic within the circle of convergence, and can be differentiated or integrated term by term within its circle of convergence.
Taylor Series
Taylor Expansion of Analytic Functions
If f(z)
is analytic within the region D: |z - z_0| < R
, then for any point z
in D
, we have
f(z) = \sum_{k=0}^{\infty}a_k (z - z_0)^k, ~~|z - z_0| < R
where a_k = \frac{1}{2\pi i}\oint_C \frac{f(\xi)}{(\xi - z_0)^{k+1}}d\xi = \frac{f^{(k)}(z_0)}{k!} ~~~ k=0, 1, 2, \cdots
Radius of Convergence of Taylor Series
Theorem
If an analytic function is expanded into a Taylor series centered at z_0
, then its circle of convergence is a circle centered at z_0
with a radius equal to the distance |z_0 - b| = R
between z_0
and the nearest singularity b
, and R
is the radius of convergence of the Taylor series.
Laurent Series
A series of the form \sum_{n=-\infty}^{\infty}c_n (z - z_0)^n
is called a Laurent series.
Theorem
If a function f(z)
is analytic within the annular domain R_1 \leq |z - z_0| \leq R_2
, then f(z)
can be expanded into a Laurent series within this annular domain
f(z) = \sum_{n-\infty}^{\infty}c_n(z-z_0)^n
where the coefficients are
c_n = \frac{1}{2\pi i}\oint_C \frac{f(\xi)}{(\xi - z_0)^{n+1}}d\xi, ~~~n=0, \pm1,\pm2,\cdots
Convergence of Laurent Series
Let a,b
be two adjacent singularities of f(z)
, and expand the function as a Laurent series centered at z_0
\sum_{n-\infty}^{\infty}c_n(z-z_0)^n
, then the series converges within the annular domain |a - z_0| < |z - z_0| < |b - z_0|
.
Isolated Singularities of Single-Valued Functions
If a function f(z)
is not differentiable at a point z_0
but is continuously differentiable in any neighborhood of z_0
except at z_0
, then z_0
is called an isolated singularity of f(z)
; if there are always non-differentiable points outside of z_0
in any neighborhood of z_0
, then z_0
is called a non-isolated singularity of f(z)
.
On the punctured neighborhood of an isolated singularity, the single-valued analytic function f(z)
can be expanded into a Laurent series \sum_{n-\infty}^{\infty}c_n(z-z_0)^n
. The positive power part is the analytic part of the series, and the negative power part is the principal part of the series.
Three Types of Singularities
Removable Singularity
Any of the following can serve as a necessary and sufficient condition or definition for an isolated singularity
-
The Laurent series of
f(z)
in the neighborhood of the singularityz_0
has no principal part. -
\lim_{z\rightarrow z_0} f(z) = c_0, c_0 \neq \infty
-
f(z)
is bounded in the neighborhood ofz_0
Pole of Order m
If the principal part of the Laurent series in the annular domain 0 < |z - z_0| < R
is a finite number of terms, i.e.
f(z) = \sum_{n=-m}^{\infty}c_n (z-z_n)^n
then z_0
is called a pole of order m, and \lim_{z\rightarrow z_0}f(z) = 0
- Laurent expansion
f(z) = \frac{1}{(z-z_0)^m \phi(z)}
,\phi(z)
is analytic and\phi(z_0) \neq 0
\lim_{z\rightarrow z_0} (z-z_0)^mf(z) = a ~(a \neq 0)
Essential Singularity
If the principal part of the Laurent expansion is an infinite number of terms, then z_0
is called an essential singularity.
- The principal part of the Laurent expansion is infinite.
- The limit
\lim_{z\rightarrow z_0} f(z)
does not exist.
Chapter 4 Residue Theorem and Its Applications
Residue Theorem
Residues at Finite Points
If the function f(z)
is expanded into a Laurent series in the neighborhood of its singularity z_0
f(z) = \sum_{n=-\infty}^{\infty}c_n(z-z_0)^n
According to the conclusion
\oint_C \frac{1}{(z-z_0)^n}dz = \begin{cases}2\pi i, n = 1 \\0, ~~n\neq 0\end{cases}
Therefore
\oint_C f(z) dz = 2\pi i c_{-1}
c_{-1}
is called the residue of f(z)
at the finite point z_0
, denoted as Resf(z_0)
or Res[f(z), z_0]
or Resf(z)|_{z=z_0}
, that is
Resf(z_0) = \frac{1}{2\pi i}\oint_C f(z) dz = c_{-1}
The residue can be obtained by calculating the Laurent series or the contour integral.
Residue at Infinity
Resf(\infty) = \frac{1}{2\pi i}\oint_Lf(z)dz
where L
is a closed contour that encircles z=0
in a clockwise direction and contains all the finite singularities of f(z)
.
At the same time, expand f(z)
into a Laurent series in the neighborhood of z=\infty
f(z) = \sum_{k=-\infty}^{\infty}C_kz^k~~~~R < |z| < \infty
Integrate both sides along the contour L
and use the conclusion from the previous section to have
\oint_L f(z) dz = C_{-1} 2\pi i
Therefore
Resf(\infty) = -C_{-1}
Theorem 1
If lim_{z\rightarrow \infty} f(z) = 0
, then the residue at infinity
Resf(\infty)=-\lim_{z\rightarrow \infty} [z \cdot f(z)]
Theorem 2
If lim_{z\rightarrow \infty} f(z) \neq 0
, then
Res[f(z), \infty] = -Res(f(\frac{1}{t}) \cdot \frac{1}{t^2}, 0)
Residue Theorem
Let the function f(z)
be analytic in the region D
except for a finite number of singularities z_1, z_2, \cdots, z_n
, and let L
be a positively oriented simple closed curve that encloses all the singularities, then
\oint_L f(z) dz = 2\pi i \sum_{k=1}^{n}Resf(z_k)
Residue Sum Theorem
Let the function f(z)
be analytic on the extended complex plane except for a finite number of singularities z_k (k=1, 2, \cdots, n)
and z=\infty
, then
\sum_{k=1}^{n}Resf(z_k) + Resf(\infty) = 0
Calculation of Residues
- When
z_0
is a removable singularityResf(z_0) = 0
- When
z_0
is an essential singularity, there is hardly any good method, only Laurent expansion or contour integration can be used. - When
z_0
is a simple polec_{-1} = \lim_{z\rightarrow z_0} (z-z_0)f(z)
- When
z_0
is a pole of order mResf(z_0) = \frac{1}{(m-1)!}\lim_{z\rightarrow z_0}\frac{d^{m-1}}{dz^{m-1}}[(z - z_0)^mf(z)]
- When
Calculation of Real Integrals Using the Residue Theorem
Type \mathop{I}
Real Integral Calculation
\int_0^{2\pi} R(cos\theta, sin\theta) d\theta = \oint_{|z|=1}R(\frac{z + z^{-1}}{2}, \frac{z - z^{-1}}{2i}) \frac{1}{iz} dz
Type \mathop{II}
Real Integral Calculation
Let f(z) = \frac{P(z)}{Q(z)}
be a rational function, where P(z)
and Q(z)
are coprime polynomials, and Q(z)
has at least twice the degree of P(z)
, i.e., \lim_{z\rightarrow \infty} zf(z) = 0
and Q(z)
has no zeros on the real axis, and there are a finite number of zeros in the upper half-plane, then
\int_{-\infty}^{+\infty}\frac{P(x)}{Q(x)} dx = 2\pi i \sum_{Imz_k > 0} Res[\frac{P(z)}{Q(z)}, z_k]
Type \mathop{III}
Real Integral Calculation
Jordan's Lemma
Let C_R
be the upper semicircle of |z|=R
, and the function f(z)
is continuous on C_R
and \lim_{z\rightarrow \infty} f(z)=0
, then
\lim_{|z|=R \rightarrow \infty} \int_{C_R}f(z)e^{iaz}dz = 0, ~~~a > 0
For the integral \int_{-\infty}^{+\infty} f(x)e^{iax}dx, ~(a > 0)
, if we take the function F(z)=f(z)e^{iaz}
, and it satisfies 1. The function F(z)=f(z)e^{iaz}
is analytic everywhere in the z
-plane except for a finite number of singularities z_1, z_2, \cdots, z_n
, and there are no singularities on the real axis 2. f(x)
is a rational function, and the degree of the denominator is at least one higher than the degree of the numerator
\int_{-\infty}^{+\infty}f(x)e^{ia}dx = 2\pi i \sum_{Imz_k > 0} Res[f(z), z_k]
Chapter 5 Fourier Series
Fourier Expansion of Periodic Functions
Fourier Series
Using the orthogonality of the trigonometric function family, we can obtain the Fourier series
f(x) = a_0 + \sum_{n=1}^{\infty}a_ncos\frac{n\pi x}{l} + b_nsin\frac{n\pi x}{l} = a_0 + \sum_{n=1}^{\infty}a_ncos\frac{2n\pi x}{T} + b_nsin\frac{2n\pi x}{T}
where
\begin{aligned}
\begin{cases}
a_0 = \frac{1}{2l} \int_{-l}^{l} f(x) dx = \frac{1}{T} \int_{-T}^{T} f(x) dx\\ \\
a_n = \frac{1}{l} \int_{-l}^{l}f(x)cos(\frac{n \pi x}{l}) dx = \frac{2}{T} \int_{-T}^{T}f(x)cos(\frac{2n \pi x}{T}) dx \\\\
b_n = \frac{1}{l} \int_{-l}^{l}f(x)sin(\frac{n \pi x}{l}) dx = \frac{2}{T} \int_{-T}^{T}f(x)sin(\frac{2n \pi x}{T}) dx\\\\
\end{cases}
\end{aligned}
Convergence of Fourier Series
Dirichlet's Theorem
If the function f(x)
is continuous everywhere, or has only a finite number of first-kind discontinuities in each period; there are only a finite number of extreme points in each period, then the series converges, and at the convergence points, we have
a_0 + \sum_{n=1}^{\infty}a_ncos\frac{n\pi x}{l} + b_nsin\frac{n\pi x}{l} = f(x)
At the points of discontinuity, we have
a_0 + \sum_{n=1}^{\infty}a_ncos\frac{n\pi x}{l} + b_nsin\frac{n\pi x}{l} = \frac{1}{2}[f(x + 0) + f(x - 0)]
Fourier Expansion of Odd and Even Functions
Fourier Expansion of Bounded Functions
- Odd extension
- Even extension
- Periodic extension
Complex Form of Fourier Series
Take the complex exponential orthogonal function family
\cdots, e^{-i\frac{k\pi x}{l}}, \cdots, e^{-i\frac{2\pi x}{l}}, e^{-i\frac{\pi x}{l}}, 1, e^{i\frac{1\pi x}{l}}, e^{i\frac{2\pi x}{l}}, \cdots, e^{i\frac{k\pi x}{l}}, \cdots
It is not difficult to derive the generalized Fourier series based on this function family
f(x) = \sum_{k=-\infty}^{\infty}C_ke^{i\frac{k\pi x}{l}}
where
C_k = \frac{1}{2l}\int_{-l}^{l}f(x)(e^{i\frac{k\pi x}{l}})^{*} dx = \frac{1}{2l}\int_{-l}^{l}f(x)e^{-i\frac{k\pi x}{l}} dx
It can be seen that C_{-k} = C^*_k
, |C_k| = \frac{\sqrt{a_k^2 + b_k^2}}{2}
Chapter 6 Mathematical Physics Boundary Value Problems
Typical Mathematical Equations
Wave Equation
u_{tt} - a^2\nabla^2u = f(x, y, z, t)
\nabla^2A = \nabla \times \nabla \times A + \nabla(\nabla \cdot A)
Heat Conduction Equation
u_t -D\nabla^2u = f(x, y, z, t)
Poisson Equation
\nabla^2u = f(x, y, z, t)
Boundary Conditions
Initial Conditions
Boundary Conditions
Dirichlet Condition
u(x, y, z, t) |_{x_0, y_0, z_0} = f(x_0, y_0, z_0, t)
Neumann Condition
\frac{\partial \vec u}{\partial \vec n}|_{x_0, y_0, z_0} = f(x_0, y_0, z_0, t)
Mixed Boundary Condition
(u + H\frac{\partial \vec u}{\partial \vec n})|_{x_0, y_0, z_0} = f(x_0, y_0, z_0, t)
Natural Boundary Condition
Second-Order Linear Partial Differential Equations
A(x, y)\frac{\partial^2u}{\partial x^2} + B(x,y)\frac{\partial^2u}{\partial x \partial y} + C(x, y)\frac{\partial^2u}{\partial y^2} + D(x, y)\frac{\partial u}{\partial x} + E(x,y)\frac{\partial u}{\partial y} + F(x, y)u = G(x,y)
Take the discriminant \Delta = B(x, y)^2 - 4A(x,y)C(x,y)
If \Delta > 0
the equation is hyperbolic, typically the wave equation u_{tt} - a^2u_{xx} = 0
If \Delta < 0
the equation is elliptic, typically the Laplace equation \nabla^2u=0
If \Delta = 0
the equation is parabolic, typically the transport equation u_t - a^2u_{xx} = 0
Linear partial differential equations can use the principle of superposition.
Method of Traveling Waves and d'Alembert's Formula
d'Alembert's Formula
For the boundary value problem
\begin{aligned}
\begin{cases}
u_{tt} - a^2u_{xx} = 0 \\
u(x, 0) = \varphi (x) \\
u_t(x, 0) = \psi (x) \\
\end{cases}
\end{aligned}
The equation can be written as
(\frac{\partial}{\partial t} + a\frac{\partial}{\partial x})(\frac{\partial}{\partial t} - a\frac{\partial}{\partial x})u = 0
After transformation
\xi = x + at ~~~~~~
\eta = x - at
It is not difficult to solve
u(x, t) = \frac{1}{2}[\varphi(x + at) + \varphi(x - at)] + \frac{1}{2a}\int_{x-at}^{x+at} \psi (\xi) d\xi
This is the solution to the one-dimensional wave equation under given initial conditions, known as d'Alembert's formula.
D'Alembert's formula is only applicable to homogeneous wave equation problems without boundaries in space.
Chapter 7 Method of Separation of Variables
Conditions for Separation of Variables
- Constant coefficient second-order homogeneous partial differential equations can always separate variables.
- Variable coefficient second-order homogeneous partial differential equations need to meet certain conditions.
- Boundary conditions need to be homogeneous.
General Steps of the Method of Separation of Variables
Variable Separation
Solve the Eigenvalue Problem
Find Particular Solutions and Superpose to Obtain the General Solution
Determine the Undetermined Coefficients Through Non-Homogeneous Initial or Boundary Conditions
General Methods for Homogenizing Non-Homogeneous Boundary Conditions
For the boundary value problem
\begin{aligned}
\begin{cases}
\frac{\partial^2u}{\partial t^2} - a^2\frac{\partial^2 u}{\partial x^2} = f(x, t) , ~~~~~~~~~~~~~~~~~~~~~~0 < x < l, t > 0\\\\
u(0, t) = g_1(t), ~u(l, t) = g_2(t), ~~~~~~~~~ t > 0 \\\\
u(x, 0) = \phi(x), ~u_t(x, 0) = \psi(x), ~~~~~ 0 < x < l\\
\end{cases}
\end{aligned}
Using the principle of superposition, let
u(x, t) = v(x, t) + w(x, t)
We can choose
w(x, t) = \frac{x}{l}[g_2(t) - g_1(t)] + g_1(t)
Then the boundary value problem for v(x, t)
\begin{aligned}
\begin{cases}
\frac{\partial^2v}{\partial t^2} - a^2\frac{\partial^2 v}{\partial x^2} = f(x, t) - \frac{x}{l}[g''_2(t) - g''_1(t)] - g''_1(t) \\\\
v(0, t) 0, ~v(l, t) = 0 \\\\
v(x, 0) = \phi(x) - \frac{x}{l}[g_2(0) - g_1(0)] - g_1(0) \\\\
v_t(x, 0) = \psi(x) - \frac{x}{l}[g'_2(0) - g'_1(0)] - g'_1(0)
\end{cases}
\end{aligned}
Common Eigenvalues and Eigenfunctions of Eigenvalue Problems
Boundary Conditions | Eigenvalues \lambda |
Eigenfunctions X(x) |
---|---|---|
X\lvert_{x=0} = 0, X\lvert_{x=l} = 0 |
\lambda = \left( \frac{n \pi}{l} \right)^2, n=1,2,3,\ldots |
\sin \left( \frac{n \pi}{l} x \right) |
X'\lvert_{x=0} = 0, X'\lvert_{x=l} = 0 |
\lambda = \left( \frac{n \pi}{l} \right)^2, n=0,1,2,\ldots |
\cos \left( \frac{n \pi}{l} x \right) |
X\lvert_{x=0} = 0, X'\lvert_{x=l} = 0 |
\lambda = \left( \frac{(n+\frac{1}{2}) \pi}{l} \right)^2, n=0,1,2,\ldots |
\sin \left( \frac{(n+\frac{1}{2}) \pi}{l} x \right) |
X'\lvert_{x=0} = 0, X\lvert_{x=l} = 0 |
\lambda = \left( \frac{(n+\frac{1}{2}) \pi}{l} \right)^2, n=0,1,2,\ldots |
\cos \left( \frac{(n+\frac{1}{2}) \pi}{l} x \right) |
Chapter 8 Series Solution and Eigenvalue Problem of Second-Order Ordinary Differential Equations
Separation of Variables in Common Orthogonal Coordinate Systems
Separation of Variables in Laplace's Equation
Spherical Coordinate System
\frac{1}{r^2}\frac{\partial}{\partial r}(r^2\frac{\partial u}{\partial r}) + \frac{1}{r^2sin^2\theta}(sin\theta \frac{\partial u}{\partial \theta}) + \frac{1}
{r^2sin^2\theta}\frac{\partial^2 u}{\partial \varphi^2} = 0
The solution to the equation is
\Phi_m = A_msinm\varphi + B_mcosm\varphi
R(r) = Cr^l + \frac{D}{r^{l+1}}
(1-x^2)\frac{d^2\Theta}{dx^2} - 2x\frac{d\Theta}{dx} + [l(l+1) - \frac{m^2}{1-x^2}]\Theta = 0
The last equation is the l
-th order associated Legendre equation, and for all physical problems, \Theta(\theta)
satisfies the following eigenvalue problem:
\begin{aligned}
\begin{cases}
(1-x^2)\frac{d^2\Theta}{dx^2} - 2x\frac{d\Theta}{dx} + [l(l+1) - \frac{m^2}{1-x^2}]\Theta = 0 \\\\
\Theta(0) = \text{finite}, ~~\Theta(\pi) = \text{finite}
\end{cases}
\end{aligned}
Cylindrical Coordinate System
\frac{1}{\rho}\frac{\partial}{\partial \rho}(\rho \frac{\partial u}{\partial \rho}) + \frac{1}{\rho^2}\frac{\partial^2 u}{\partial \varphi^2} + \frac{\partial^2u}{\partial z^2} = 0
The solution for \Phi
is
\Phi(\varphi) = A_msinm\varphi + B_mcosm\varphi
When \mu = 0
, the solution is
\begin{aligned}
\begin{cases}
R(\rho) = \begin{cases}A_0 + B_0ln\rho ,~~m=0\\\\A_m\rho^m + B_m\rho^{-m}, ~~ m\neq 0\end{cases}\\\\
Z(z) = C + Dz
\end{cases}
\end{aligned}
When \mu > 0
, the solution is
Z(z) = Ce^{\sqrt{u}z} + De^{-\sqrt{u}z}
The equation for R
is
x^2\frac{d^2R}{dx^2} + x\frac{dR}{dx} + (x^2 - m^2)R = 0
This is the standard Bessel equation.
When \mu < 0
, the solution is
Z(z) = Csin{\sqrt{-\mu}z} + Dcos{\sqrt{-\mu}z}
The equation for R
is
x^2\frac{d^2R}{dx^2} + x\frac{dR}{dx} + (-x^2 - m^2)R = 0
This is the standard imaginary Bessel equation.
Separation of Variables in the Wave Equation
u_{tt}(\vec{r}, t) - a^2\nabla^2u(\vec{r}, t) = 0
The result is
\begin{aligned}
\begin{cases}
T''(t) + k^2a^2T(t) = 0 \\\\
\nabla^2V(\vec{r}) + k^2V(\vec{r}) = 0
\end{cases}
\end{aligned}
Separation of Variables in the Heat Conduction Equation
u_{t}(\vec{r}, t) - a^2\nabla^2u(\vec{r}, t) = 0
The result is
\begin{aligned}
\begin{cases}
T'(t) + k^2a^2T(t) = 0 \\\\
\nabla^2V(\vec{r}) + k^2V(\vec{r}) = 0
\end{cases}
\end{aligned}
Separation of Variables in the Helmholtz Equation
Spherical Coordinate System
\frac{1}{r^2} \frac{\partial}{\partial r} \left( r^2 \frac{\partial \Psi}{\partial r} \right) + \frac{1}{r^2 \sin\theta} \frac{\partial}{\partial \theta} \left( \sin\theta \frac{\partial \Psi}{\partial \theta} \right) + \frac{1}{r^2 \sin^2\theta} \frac{\partial^2 \Psi}{\partial \phi^2} + k^2 \Psi = 0
The solution is
\Phi_m = A_msinm\varphi + B_mcosm\varphi
The equation for R
is
r^2\frac{d^2R}{dr^2} + 2r\frac{dR}{dr} + [k^2r^2 - l(l+1)]R = 0
Let x = kr
, R(r) = \sqrt{\frac{\pi}{2x}}y(x)
x^2\frac{d^2y}{dx^2} + x\frac{dy}{dx} + (x^2 - (l + \frac{1}{2})^2)y = 0
Cylindrical Coordinate System
\frac{1}{r} \frac{\partial}{\partial r} \left( r \frac{\partial \Psi}{\partial r} \right)+ \frac{1}{r^2} \frac{\partial^2 \Psi}{\partial \phi^2}+ \frac{\partial^2 \Psi}{\partial z^2}+ k^2 \Psi = 0
Series Solution in the Neighborhood of Regular Points
Regular and Singular Points of the Equation
For the equation
\frac{d^2w(z)}{dz^2} + p(z)\frac{dw(z)}{dz} + q(z)w(z) = 0
If the functions p(z)
and q(z)
are analytic at z_0
, then z_0
is a regular point of the equation. If z_0
is a singular point of the equation, then z_0
is a singular point of the equation. If z_0
is a singular point of p(z)
of no higher than the first order, and q(z)
is no higher than the second order, then z_0
is called a regular singular point of the equation.
Power Series Solution in the Neighborhood of a Regular Point
w(z) = \sum_{k=0}^{\infty}a_k(z-z_0)^k
Sturm-Liouville Eigenvalue Problem
The Sturm-Liouville Equation
\begin{aligned}
\frac{d}{dx}[k(x)\frac{dy}{dx}] - q(x)y + \lambda P(x)y = 0 \\\\
k(x) \geq 0, ~~q(x) \geq 0, ~~P(x) \geq 0, ~~ a\leq x \leq b
\end{aligned}
Chapter 9 Special Functions
Legendre Polynomials
Series Solution of the Legendre Equation
The Legendre equation in spherical coordinates is:
\frac{1}{\sin\theta}\frac{d}{d\theta}(\sin\theta \frac{d\Theta}{d\theta}) + [l(l + 1) - \frac{m^2}{\sin^2\theta}]\Theta = 0
This equation can be written as:
(1-x^2)\frac{d^2y}{dx^2} - 2x\frac{dy}{dx} + [l(l+1) - \frac{m^2}{1-x^2}]y = 0
The solution to the Legendre equation is the Legendre series:
P_l(x) = \sum_{k=0}^{[\frac{l}{2}]}(-1)^k\frac{(2l - 2k)!}{2^lk!(l-k)!(l-2k)!}x^{l-2k}
The first few terms are:
P_0(x) = 1
P_1(x) = x
P_2(x) = \frac{1}{2}(3x^2 - 1)
P_3(x) = \frac{1}{2}(5x^3 - 3x)
P_4(x) = \frac{1}{8}(35x^4 - 30x^2 + 3)
Properties of Legendre Polynomials
Parity
P_l(-x) = (-1)^lP_l(x)
Special Values
P_{2n+1}(0)=0
P_{2n} = (-1)^n\frac{(2n-1)!!}{(2n)!!}
P_l(1) = 1, P_l(-1) = (-1)^l
Orthogonality
Eigenfunctions of the S-L problem are weighted orthogonal.
Differential and Integral Representations of Legendre Polynomials
P_l(x) = \frac{1}{2^ll!}\frac{d^l}{dx^l}(x^2 - 1)^l
P_l(x) = \frac{1}{2\pi i}\frac{1}{2^l}\oint_C\frac{(\xi^2-1)^l}{(\xi-x)^{l+1}}dx
P_l^m(x) = \frac{(1-x^2)^{\frac{m}{2}}}{2^ll!}\frac{d^{l+m}}{dx^{l+m}}(x^2-1)^l
Boundary Value Problem
Solution of Laplace's Equation in Spherical Coordinates
\begin{aligned}
\begin{cases}
R_l(r) = C_lr^l + \frac{D_l}{r^{l+1}} \\\\
\Phi_m(\varphi) = Asinm\varphi + Bcosm\varphi \\\\
\Theta_{m, l}(\theta) = P_l^m(\cos\theta)
\end{cases}
\end{aligned}
Recurrence Relations for Legendre Polynomials
From the generating function for Legendre polynomials, we derive the following recurrence relation:
(n+1)P_{n+1}(x) - (2n+1)xP_n(x) + nP_{n-1}(x) = 0
P_n(x) = P'_{n+1}(x) - 2xP'_n(x) + P'_{n-1}(x), ~~ n\geq1
From these two formulas, we can derive:
(2n+1)P_n(x) = P'_{n+1}(x) - P'_{n-1}(x), ~~n \geq 1
P'_{n+1}(x) = (n+1)P_n(x) + xP'_n(x), ~~~n \geq 0
nP_n(x) = xP'_n(x) - P'_{n-1}(x), ~~~n\geq1
(x^2-1)P'_n(x) = nxP(x) - nP_{n-1}(x), ~~n\geq1
Bessel Functions
Three Types of Cylindrical Functions
The Bessel equation is:
x^2\frac{d^2y}{dx^2} + x\frac{dy}{dx} + (x^2 - \nu^2)y = 0
The general solution to the Bessel equation is:
y(x) = c_1y_1(x) + c_2y_2(x)
where
y_1(x) = \sum_{n=0}^{\infty}\frac{(-1)^nc_0\Gamma(\nu + 1)}{2^{2n}n!\Gamma(\nu + n + 1)}
y_2(x) = \sum_{n=0}^{\infty}\frac{(-1)^nc_0\Gamma(-\nu + 1)}{2^{2n}n!\Gamma(-\nu + n + 1)}
By choosing the constant c_0 = \frac{1}{2^{\nu}\Gamma(\nu + 1)}
in y_1(x)
, we define y_1(x)
as J_\nu(x)
, known as the \nu
th-order Bessel function:
J_\nu(x) = \sum_{n=0}^{\infty}\frac{(-1)}{n!\Gamma(\nu + n + 1)}\left(\frac{x}{2}\right)^{2n+\nu}
By choosing the constant c_0 = \frac{1}{2^{-\nu}\Gamma(\nu + 1)}
in y_2(x)
, we define y_2(x)
as J_{-\nu}(x)
, known as the \nu
th-order Bessel function:
J_{-\nu}(x) = \sum_{n=0}^{\infty}\frac{(-1)}{n!\Gamma(-\nu + n + 1)}\left(\frac{x}{2}\right)^{2n-\nu}
J_\nu(x)
and J_{-\nu}(x)
are called the first kind cylindrical functions. When \nu
is not an integer, they are linearly independent. In this case, the general solution is:
y(x) = A_\nu J_\nu(x) + B_\nu J_{-\nu}(x)
When \nu
is an integer, the second general solution is:
y(x) = A_\nu J_\nu(x) + B_\nu N_{\nu}(x)
where
N_{\nu}(x) = \frac{\cos\nu\pi J_{\nu}(x) - J_{-\nu}(x)}{\sin\nu \pi}
N_{\nu}(x)
is called the second kind cylindrical function, also known as the Neumann function.
The third form of the general solution is:
y(x) = A_\nu H_\nu^{(1)}(x) + B_\nu H_{\nu}^{(2)}(x)
where
\begin{aligned}
\begin{cases}
H_{\nu}^{(1)}(x) = J_{\nu}(x) + iN_{\nu}(x)\\\\
H_{\nu}^{(2)}(x) = J_{\nu}(x) - iN_{\nu}(x)
\end{cases}
\end{aligned}
These are called the third kind cylindrical functions, also known as Hankel functions.
Properties of Bessel Functions
Values at Special Points
\begin{aligned}
\lim_{x\rightarrow0}J_0(x) = 1, \lim_{x\rightarrow0}J_\nu(x) = 0, \lim_{x\rightarrow0}J_{-\nu}(x) = \infty, \lim_{x\rightarrow0}N_\nu(x) = \infty, \\\\
\lim_{x\rightarrow\infty}J_\nu(x) = 0, \lim_{x\rightarrow\infty}J_{-\nu}(x) = 0, \lim_{x\rightarrow\infty}N_\nu(x) = 0
\end{aligned}
Recurrence Relations for Bessel Functions
\frac{d}{dx}[x^vZ_v(x)] = x^vZ_{v-1}(x)
\frac{d}{dx}[x^{-v}Z_v(x)] = -x^{-v}Z_{v+1}(x)
Common Corollaries
J'_0(x) = -J_1(x)
[xJ_1(x)]' = xJ_0(x)
xJ_1(x) = \int_0^x\xi J_0(\xi) d\xi