Ch11 14 PDF
Ch11 14 PDF
Ch11 14 PDF
2016年9月19日 下午 06:33
微分方程式
Chapter 11-14
管傑雄
台大電機系
Chapter 11 Orthogonal Functions and Fourier
Series
Orthogonal Functions
If the inner product of two functions is zero, i.e.,
=0
then f1 and f2 are said to be orthogonal.
Ch11-14 第 1 頁
Example: f1(x) = x2 and f2(x) = x3 are orthogonal on [-1, 1] since
Orthogonal Sets
A set of real-valued functions
{0(x), 1(x), 2(x), .....}
is said to be orthogonal on an interval [a, b] if
Orthonormal set
If {n(x)} is an orthogonal set of functions on [a, b] with the properties that ||n(x)|| = 1
for n = 0, 1, 2, .... , then {n(x)}is said to be an othonormal set on the interval.
Example: {1, cosx, cos2x, .... } are an othogonal set on [-, ] since
,m n
The norms of each function in the set are
||0|| = 2
||m|| = for m 0
Hence, we can form an othonormal set as follows
Therefore,
Complete Sets
The main concern of the possibility of the generalized Fourier expansion of f(x) with
respect to {n(x)} is the completeness of {n(x)}. If f(x) is orthogonal to every n(x), this
indicates that {n(x)} is not complete. In a complete orthogonal set of functions, the
generalized Fourier expansion of a function is always possible. This also means that the
only function orthogonal to each member of the set is the zero function.
Note: The condition of completeness: Since the generalized Fourier expansion is true for any
function f(x) on [a, b],
Ch11-14 第 2 頁
we can conclude that
for m n
The usual assumption is that w(x) > 0 on the interval of othogonality [a, b].
for n > 0
for n > 0
then the cosfficients can be determined in exactly the same manner as in the discussion of
Generalized Fourier series.
Ch11-14 第 3 頁
The trigonometric series is said to be the Fourier series of the function f and the coefficients
are referred to as Fourier coefficients of f.
Example 1: Expand
in a Fourier series.
Solution:
Therefore,
Periodic Extension
A Fourier series not only represents the function on the interval [-p, p] but also gives
the periodic extension of f outside this interval. At the end points of x = -p and x = p,
the Fourier series will converges to
Example 2: For the function in example 1, at 0, 2, 4, ..., the series converges to /2 and at
, 3, ... , the series converges to zero.
Ch11-14 第 4 頁
Properties of even and odd functions:
(1) The product of two even functions is even
(2) The product of two odd functions is even
(3) The product of an even function and an odd function is odd
(4) The sum (difference) of two even functions is even
(5) The sum (difference) of two odd functions is odd
(6) If f is even, then
where
(2) The Fourier series of an odd function on the interval (-p, p) is the sine series
Thus
Example 2: Expand
in a Fourier series.
Ch11-14 第 5 頁
in a Fourier series.
Solution:
Therefore,
The graph of S15 has pronounced spikes near the discontinuities at x = 0, , and so on. This
overshooting behavior near discontinuous points of f is known as Gibbs phenomenon.
Half-Range Expansions
If f(x) is defined on the interval 0 < x < L, then we can expand the functions in three
ways:
(1) Cosine series with period 2L
(2) Sine series with period 2L
(3) Fourier series with period L
where (1) and (2) are known as half-range expansion.
(1) (2) (3)
Ch11-14 第 6 頁
Example 6 Expand f(x) = x2, 0 < x < L (a) in a cosine series, (b) in a sine series, and (c) in a
Fourier series.
Solution:
(a)
Hence,
(b)
Hence
(c)
Hence,
Ch11-14 第 7 頁
Periodic Driving Force
We consider the differential equation
Example: Suppose m = 1/16 and k = 4 and f(t) is shown in the figure. Please find a particular
solution.
Solution:
i.e.,
Ch11-14 第 8 頁
Hence,
We find no n such that Bn = 0. In general, if thereis a value of n, say N, for which N/p = ,
where
where A, B, C, ... , G are functions of x and y. When G(x, y) = 0, the equation is said to be
homogeneous; otherwise it is nonhomogeneous.
Separation of Variables
By seeking a particular solution, we suppose
u(x, y) = X(x)Y(y)
.
Solution: Suppose
u(x, y) = X(x)Y(y)
then
Hence,
u = A1e ycosh2x + B1e ysinh2x
2 2
Ch11-14 第 9 頁
X = c4cos2x + c5sin2x and Y = c6e- y 2
Hence,
u = A2e- ycos2x + B2e- ysin2x
2 2
Note: Separation of variables is not a general method for finding particular solutions; some
linear partial differential equations are simply not separable.
Superposition Princilple
If u1, u2, ... , uk are solutions of a homogeneous linear partial differential equation, then
the linear combination
u = c1u1 + c2u2 + ... + ckuk
where ci, i = 1, 2, ... , k, are constants, is also a solution.
Note: General solution的重疊原理僅適用於Homogeneous case;但不適用nonhomogeneous
case。同理,若有boundary or initial conditions存在,也只是適用邊界或初始條件為零的情
形。
Classification of Equations
The linear second-order partial differential equation
, k>0
One-dimensional Wave Equation:
Heat Equation:
Assumptions:
1. The flow of heat within the rod takes place only in the x-direction.
2. The lateral, or curved, surface of the rod is insulated; that is, no heat escapes from this
surface.
3. No heat is generated within the rod.
4. The rod is homogeneous; that is, the mass per unit volume is a constant.
Ch11-14 第 10 頁
4. The rod is homogeneous; that is, the mass per unit volume is a constant.
5. The specific heat and thermal conductivity K of the material of the rod are constants.
Define the temperature u(x, t) at the position x of the rod and time t. Then the quantity of heat
in the element x (relative to the temperature zero) at the position x is
Q = Axu
The heat energy increasing rate is
Qt = Axut
The quantity is equal to the heat energy increasing rate due to the conduction
-KAux(x, t) - (-KAux(x + x, t))
Therefore, we get the partial differential equation
Wave Equation
Assumptions:
1. The string is perfectly flexible
2. The string is hopmogeneous; that is, its mass per unit length is a constant.
3. The displacements u are small compared to the length of the string.
4. The slope of the surve is small at all points.
5. The tension T of the string is constant.
6. The tension is large compared with the force of gravity.
7. No other external forces act on the string.
Laplace's Equation
As in electrostatic, gravitational, and velocity potential in fluid mechanics.
Initial Conditions
1. First partial derivative of t: Single initial condition
Ch11-14 第 11 頁
1. First partial derivative of t: Single initial condition
u(x, 0) = f(x), 0 < x < L
2. Second partial derivative of t: two initial conditions
u(x, 0) = f(x),
, 0<x<L
Boundary Conditions
We have to specify the boundary conditions at all boundary points at any time t to solve
the problem. In general, there are three types of boundary conditions:
1. Dirichlet condition: u
2. Neumann condition:
3. Robin condition:
, where h is a constant.
Boundary-Value Problem:
The condition to solve the problem: Initial conditions at all points and Boundary
conditions at all time.
For example:
Solve:
, 0<x<L
u(0, t) = 0, u(L, t) = 0, t > 0
Solve:
, 0<y<b
u(x, 0) = 0, u(x, b) = f(x), 0 < x <a
Modifications
Ch11-14 第 12 頁
For example, if there is heat transfer from the lateral surface of a rod into a surrounding
medium that is held at a constant temperature u0, then the heat equation becomes
For the wave equation, if there are an external force, a damping force, and a restoring force,
then the equation becomes
, 0<x<L
Solution: Seoaratiing variables gives
so that
X" + 2X = 0
T" + 2a2T = 0
and therefore
X = c1cosx + c2sinx
T = c3cosat + c4sinat
The boundary condition gives
c1 = 0 and
and
Setting t = 0 gives
Ch11-14 第 13 頁
Hence,
Among them, f1 is called the fundamental frequency or the first harmonic, and fn (n 1) is
called overtone. The one of n = 2 is the first overtone.
, 0<y<b
Ch11-14 第 14 頁
Solution: Separating variables leads to
X" + 2X = 0
Y" - 2Y = 0
and
X = c1cosx + c2sinx
Y = c3coshx + c4sinhx
For = 0, we have
X = c1 + c2x
Y = c3 + c4y
For case, the solutions are
2
X = c1coshx + c2sinhx
Y = c3cosx + c4sinx
Under the first three boundary conditions, the third set of solutions is zero, and the second set
is
XY = A0y
The first set of the solutions is
, n = 1, 2, 3, ...
The superposition principle yields another solution is
Substituting y = b gives
This leads to
Ch11-14 第 15 頁
Solution: Separating the variables gives
where is a number greater than zero. Hence, the solution forms we should choose is
X = c1sinx
Y = c2e-y
The boundary conditions at x = 0 and x = a leads to
, n = 1, 2, 3, ...
The series solution form is
That is,
Ch11-14 第 16 頁
With complex numbers, we can write the solution as
If we define
Z = e(i/a)(x+iy)
then the solution is written as
We consider that
Then
Therefore,
Dirichlet Problem
A boundary-value problem in which we seek a solution to an elliptic partial differential
equation, such as Laplace's equation within a region such that u takes on prescribed
Ch11-14 第 17 頁
equation, such as Laplace's equation within a region such that u takes on prescribed
valuues on the entire boundary of the region is called a Dirichlet problem.
Superposition Pronciple
A Dirichlet problem for a retangle can be solved readily by separation of variables
when homogeneous boundary conditions are specified on two parallel boundaries.
However, the method of separation of variables is not applicable to a Dirichlet problem
when the boundary conditions on all four sides of the retangle are not homogeneous. To
get around this difficuility we break the problem
Problem I
Problem II:
The solution of the original problem is u(x, y) = u1(x, y) + u2(x, y), which can be proved by
the superposition principle.
A solution of problem I is
where
A solution of probblem II is
Ch11-14 第 18 頁
Chapter 14 Integral Transform Method
14.3 & 14.4 Fourier Integral and Transform
Fourier Integral
From the definition of the Fourier seriers,
we define n = n/p, = n+1 - n = /p. Then the function f(x) can be written as
If we define
then
Ch11-14 第 19 頁
Note: Convergence at discontinuity of the above integral
Complex Form
Ch11-14 第 20 頁
Fourier Transform Pairs
(i) Fourier Transform
Example: Represent f(x) = e-x, x > 0, (a) by a cosine integral and (b) by a sine integral.
Solution: Using the integration by parts, we find
and
Ch11-14 第 21 頁
Inversion Transform (Integral)
Existence
F{1}, Fs{1}, and Fc{1} do not exist. Sufficient conditions for existence are that f be
absolutely integrable on the appropriate interval and that f and f' be piecewise
continuous on every finite interval.
Operational Properties
(1) Fourier Transform
Suppose that f is continuous and absolutely integrable on the interval (-, ) and f' is
piecewise continuous on every finite interval. If f(x) 0 as x , then integration by parts
gives
Similarly, if f" is piecewise continuous on every finite interval, and f'(x) 0 as x , we have
2nd derivative
Note:
Fourier Transform: (-,);己知初始條件不定
Fourier Sine(己知初始條件f(0)) or Cosine(己知初始條件f’(0)) Transform: [0, )
Ch11-14 第 22 頁
, -<x<, t>0
subject to u(x,0) = f(x) where
.
Solution: Using the Fourier transform with respect to x gives
This gives
Example 2: Solve
, 0<x<
Solution: The Fourier cosine transform with respect to y gives
These leads to
c1 = 0,
That is,
Ch11-14 第 23 頁
The inversion integral gives
Ch11-14 第 24 頁
2016.12.21 (三) ok
2017年1月24日 上午 11:24
Ch11-14 第 25 頁
Orthogonal function&set
2017年1月24日 上午 11:24
Orthogonal Functions
If the inner product of two functions is zero, i.e.,
=0
then f1 and f2 are said to be orthogonal.
Orthogonal Sets
A set of real-valued functions
{0(x), 1(x), 2(x), .....}
is said to be orthogonal on an interval [a, b] if
Orthonormal set
If {n(x)} is an orthogonal set of functions on [a, b] with the properties that ||n(x)|| = 1
for n = 0, 1, 2, .... , then {n(x)}is said to be an othonormal set on the interval.
Example: {1, cosx, cos2x, .... } are an othogonal set on [-, ] since
,m n
The norms of each function in the set are
||0|| = 2
||m|| = for m 0
Hence, we can form an othonormal set as follows
Ch11-14 第 26 頁
Ch11-14 第 27 頁
Fourier Series 簡介
2017年1月24日 上午 11:27
Ch11-14 第 28 頁
Generalized Fourier Series & complete test
2017年1月24日 上午 11:27
Complete Sets
The main concern of the possibility of the generalized Fourier expansion of f(x) with
respect to {n(x)} is the completeness of {n(x)}. If f(x) is orthogonal to every n(x), this
indicates that {n(x)} is not complete. In a complete orthogonal set of functions, the
generalized Fourier expansion of a function is always possible. This also means that the
only function orthogonal to each member of the set is the zero function.
假設 orthogonal
又
比較可得
此式可作為complete檢查 (但不容易)
Ch11-14 第 29 頁
2016.12.23 (五)
2017年1月24日 上午 11:32
Ch11-14 第 30 頁
Fourier Series
2017年1月24日 上午 11:32
for n > 0
for n > 0
Suppose f is a function defined on [-p, p] that can be expanded in the trigonometric series
then the cosfficients can be determined in exactly the same manner as in the discussion of
Generalized Fourier series.
The trigonometric series is said to be the Fourier series of the function f and the coefficients
are referred to as Fourier coefficients of f.
Example 1: Expand
Ch11-14 第 31 頁
in a Fourier series.
Solution:
Therefore,
心電圖信號,用Fourier series取30項直接算,原始信號頭尾不相接
可以發現左右兩邊邊緣edge fitting出的曲線會有問題
Ch11-14 第 32 頁
使用我們方法,把原始信號頭尾相接後,還原回去可以發現很準,edge效應有改善
Ch11-14 第 33 頁
Ch11-14 第 34 頁
FCS & FSS
2017年1月27日 下午 05:15
其中 為even function
為 odd function
where
(2) The Fourier series of an odd function on the interval (-p, p) is the sine series
Ch11-14 第 35 頁
Example 1: Expand f(x) = x, -2< x < 2 in a Fourier series.
Solution:
Thus
Example 2: Expand
in a Fourier series.
Solution:
Therefore,
Ch11-14 第 36 頁
The graph of S15 has pronounced spikes near the discontinuities at x = 0, , and so on. This
overshooting behavior near discontinuous points of f is known as Gibbs phenomenon.
Ch11-14 第 37 頁
取得的畫面剪輯: 2017/1/27 下午 05:32
Ch11-14 第 38 頁
取得的畫面剪輯: 2017/1/27 下午 05:33
Ch11-14 第 39 頁
取得的畫面剪輯: 2017/1/27 下午 05:34
Ch11-14 第 40 頁
Ch11-14 第 41 頁
Gibb's phenomenon 說明 *** https://en.wikipedia.org/wiki/Gibbs_phenomenon
2017年1月27日 下午 05:44
n=1 剩餘
n==3 剩餘
n==5 剩餘
Ch11-14 第 42 頁
取得的畫面剪輯: 2017/1/27 下午 05:48
n=7 剩餘
n=9 剩餘
n=11 剩餘
Ch11-14 第 43 頁
取得的畫面剪輯: 2017/1/27 下午 05:50
n=19 剩餘
n=39 剩餘
n=59 剩餘
Ch11-14 第 44 頁
取得的畫面剪輯: 2017/1/27 下午 05:52
n=99 剩餘
Gibb's phenomenon導致傅立葉展開需要無窮多組項才收斂,實際上是沒收斂
會有Osillation產生,稱之為overshoot
Ch11-14 第 45 頁
Half range expansion
2017年1月24日 上午 11:32
Half-Range Expansions
If f(x) is defined on the interval 0 < x < L, then we can expand the functions in three
ways:
(1) Cosine series with period 2L
(2) Sine series with period 2L
(3) Fourier series with period L
where (1) and (2) are known as half-range expansion.
(1) (2) (3)
Example 6
Expand f(x) = x2, 0 < x < L (a) in a cosine series, (b) in a sine series, and (c) in a Fourier
series.
Solution:
(a)
這個方法,先利用f(-x)=f(x),將原本信號對y軸做對稱,得到-L<x<L的信號,稱之為even-
function的方式.再把這個信號往兩旁無限延伸。這個方法的好處是只要計算cosine
和dc項的係數就好,因為sine是奇函數(odd funciton),sine的係數在這種展開方式必定
是0。
Hence,
可以注意到,(2k+1)L 連接處為"折點"(一階微分開始不連續),所得到的係數收斂
速度為 ; 2kL 連接處為無窮多階可微,在係數中並沒有出現與之相關的項。
(b)
這個方法,先利用f(-x)=-f(x),將原本信號對y軸做對稱,得到-L<x<L的信號,稱之為odd-
function的方式.再把這個信號往兩旁無限延伸。這個方法的好處是只要計算sine的
係數就好,因為cosine和dc是偶函數(even funciton),cosine和dc的係數在這種展開方式
必定是0。
Hence
Ch11-14 第 46 頁
Hence
(2k+1)L 連接處為"不連續點(jump)"(零階微分開始不連續),所得到的係數
項收斂速度為 ;在 2kL 連接處在二階微分開始不連續,所得到的係數
項收斂速度為
(c)
這個方法,先利用f(x-L)=f(x) ,將原本信號往左延伸,得到-L<x<L的信號,再把這個信號
往兩旁無限延伸。
Hence,
任何函數均可分解成奇函數與偶函數相加,如圖所示。其中 偶函數項使
用cos做展開; 偶函數項使用sin 項做展開;偶函數在連接處以"折點"連
接(一階微分開始不連續),所以cos 項對應的係數 收斂速度為 ;奇函數在連接
處以"不連續點(jump)"連接(零階微分開始不連續),所以sin 項對應的係數 收斂
速度為
Ch11-14 第 47 頁
Our method ***
2017年1月24日 上午 11:33
由上述討論可以得知,
開始不連續 收斂速度
零階微分
一階微分
二階微分
三階微分
… …
Ch11-14 第 48 頁
2016.12.28 (三) ok
2017年1月24日 上午 11:33
Ch11-14 第 49 頁
PDE in Rectangular Coor.
2017年1月24日 上午 11:35
比較 :
1 independent variable 、n dependent variables -> system of DE
n independent variable 、1 dependent variables -> PDE
General form
以2元2階偏微分方程式為例
為 8-dimension 的問題
Linear Equation
The general form of a linear second-order partial differential equation in two
independent variables x and y is given by
Superposition Princilple
If u1, u2, ... , uk are solutions of a homogeneous linear partial differential
equation, then the linear combination
u = c1u1 + c2u2 + ... + ckuk
where ci, i = 1, 2, ... , k, are constants, is also a solution.
Note: General solution的重疊原理僅適用於Homogeneous case;但不適用
nonhomogeneous case。同理,若有boundary or initial conditions存在,也只是適
用邊界或初始條件為零的情形。
Classification of Equations
The linear second-order partial differential equation
Ch11-14 第 50 頁
物理應用
2017年1月24日 上午 11:36
Heat Equation:
Assumptions:
1. The flow of heat within the rod takes place only in the x-direction.
2. The lateral, or curved, surface of the rod is insulated; that is, no heat escapes from this
surface.
3. No heat is generated within the rod.
4. The rod is homogeneous; that is, the mass per unit volume is a constant.
5. The specific heat and thermal conductivity K of the material of the rod are constants.
Define the temperature u(x, t) at the position x of the rod and time t. Then the quantity of heat
in the element x (relative to the temperature zero) at the position x is
Q = Axu
The heat energy increasing rate is
Qt = Axut
The quantity is equal to the heat energy increasing rate due to the conduction
-KAux(x, t) - (-KAux(x + x, t))
Therefore, we get the partial differential equation
Wave Equation
Assumptions:
1. The string is perfectly flexible
2. The string is hopmogeneous; that is, its mass per unit length is a constant.
3. The displacements u are small compared to the length of the string.
4. The slope of the surve is small at all points.
5. The tension T of the string is constant.
6. The tension is large compared with the force of gravity.
7. No other external forces act on the string.
Ch11-14 第 51 頁
The net vertical force acting on the element s of the string is
T sin2 - T sin1 T tan1 - T tan2
= T [ux(x+x, t) - ux(x, t)]
From the Newton's second law, we get
T [ux(x+x, t) - ux(x, t)] = xutt
That is,
Ch11-14 第 52 頁
Boundary condition & initial condition
2017年1月28日 下午 04:59
Initial Conditions
1. First partial derivative of t: Single initial condition
u(x, 0) = f(x), 0 < x < L
2. Second partial derivative of t: two initial conditions
u(x, 0) = f(x),
, 0<x<L
Boundary Conditions
We have to specify the boundary conditions at all boundary points at any time t to solve
the problem. In general, there are three types of boundary conditions:
1. Dirichlet condition: u
2. Neumann condition:
3. Robin condition:
, where h is a constant.
Boundary-Value Problem:
The condition to solve the problem: Initial conditions at all points and Boundary
conditions at all time.
For example:
Solve:
, 0<x<L
u(0, t) = 0, u(L, t) = 0, t > 0
Solve:
, 0<y<b
u(x, 0) = 0, u(x, b) = f(x), 0 < x <a
Ch11-14 第 53 頁
Modifications
For example, if there is heat transfer from the lateral surface of a rod into a surrounding
medium that is held at a constant temperature u0, then the heat equation becomes
For the wave equation, if there are an external force, a damping force, and a restoring force,
then the equation becomes
Ch11-14 第 54 頁
求解 Uc
2017年1月24日 上午 11:36
Separable 變數分離
假設解可做變數分離,
代入原方程式,利用eigen value找出可以展開的基底
所使用的概念與orthogonal base的概念相近,在解空間找出一個較小的集合,使用此集合
展開整個解集合。
.
Solution: Suppose
u(x, y) = X(x)Y(y)
then
為x only ; 為y only,兩者要相等,表示僅可能為常數,再討論常數可以是正、
負、零。
Ch11-14 第 55 頁
Case I when 2 > 0, we get two equations
X" - 42X = 0 and Y' - 2Y = 0
These equations have the solutions
X = c1cosh2x + c2sinh2x and Y = c3e y 2
Hence,
u = A1e ycosh2x + B1e ysinh2x
2 2
Hence,
u = A2e- ycos2x + B2e- ysin2x
2 2
Note: Separation of variables is not a general method for finding particular solutions; some
linear partial differential equations are simply not separable.
Superposition Princilple
If u1, u2, ... , uk are solutions of a homogeneous linear partial differential equation, then the
linear combination
u = c1u1 + c2u2 + ... + ckuk
where ci, i = 1, 2, ... , k, are constants, is also a solution.
Note: General solution的重疊原理僅適用於Homogeneous case;但不適用nonhomogeneous
case。同理,若有boundary or initial conditions存在,也只是適用邊界或初始條件為零的情
形。
Ch11-14 第 56 頁
數值方法 (管老師迭代法)
2017年1月24日 上午 11:37
在x=0處展開
求得
使用泰勒展開式即可求得
Example
已知
只討論唯一解析解區,故其解U 無窮多階可微(使用泰勒
展開式所必須),因此偏微分的順序可以任意調換
……
可以用 、 來表示
Ch11-14 第 57 頁
2017.01.04 (三)
2017年1月24日 上午 11:39
Ch11-14 第 58 頁
電磁學例題
2017年1月24日 上午 11:38
where is a number greater than zero. Hence, the solution forms we should choose is
X = c1sinx
Y = c2e-y
The boundary conditions at x = 0 and x = a leads to
, n = 1, 2, 3, ...
The series solution form is
That is,
Ch11-14 第 59 頁
With complex numbers, we can write the solution as
If we define
Z = e(i/a)(x+iy)
then the solution is written as
We consider that
Then
Therefore,
可以注意到closed form
Ch11-14 第 60 頁
為不可變數分離的解,但可以使用eigen function 無窮多項將
其展開。
Ch11-14 第 61 頁
Singular point 物理上的解釋
2017年1月24日 上午 11:38
Ch11-14 第 62 頁
Eigen function 在邏輯上的缺失
2017年1月28日 下午 04:55
解微分方程,常需要使用eigen-space
eigen-space定義為由Eigen function裡面的基底所span的空
間
例如求解y''+k^2 y=0可得y=Acos(kx)+Bsin(kx)
則我們使用了cos(kx)和sin(kx)這兩個eigen-function作為
eigen-space基底,而Homogeneous 解空間為可無限多次微分
的空間
解一個問題通常需要多種類的eigen-space
此為解微分方程的多樣化邏輯
不同於傳統的單一性邏輯
Ch11-14 第 63 頁
管老師 基底
2017年1月24日 上午 11:39
Ch11-14 第 64 頁
Fourier integral transform
2017年1月24日 上午 11:40
Fourier Integral
From the definition of the Fourier seriers,
由上圖可以看出,由p所定義的區間所取到的函數圖形可能不為週期性函數(要求在左
右邊界連續、一階微分連續、…,才可以將數個圖形接起來而成為週期性函數)。若
將p往外延展,使p趨近無窮大,即可得到左右邊界連續、一階微分連續、…。
Ch11-14 第 65 頁
we define n = n/p, = n+1 - n = /p. Then the function f(x) can be written as
其中 有限,但趨近於0,因此該項為0
If we define
then
Ch11-14 第 66 頁
Complex Form
2017年1月24日 上午 11:41
為 的even function
為 的odd function
項為 的even function
項為odd * odd ,亦為 的even function
因此 為 的even function
所以可以改寫成
將下限改為負無窮大,外面乘上
將cos與sin改用exp函數取代
其中
因為上下限為正負無窮大,所以 可改為
同理
因此
Ch11-14 第 67 頁
又因為上下限為正負無窮大,所以 可改為
其中
Ch11-14 第 68 頁
FCT & FST
2017年1月24日 上午 11:41
Ch11-14 第 69 頁
Fourier transform 的微分
2017年1月28日 下午 11:25
同理
Ch11-14 第 70 頁
Fourier sin transform 與Fourier cos transform對於二階偏導數可以維持原本的轉換,在
解題上可以使用。
Ch11-14 第 71 頁
結合 FT求解DE
2017年1月24日 上午 11:40
, -<x<, t>0
subject to u(x,0) = f(x) where
.
Solution: Using the Fourier transform with respect to x gives
This gives
Example 2: Solve
, 0<x<
Solution: The Fourier cosine transform with respect to y gives
Ch11-14 第 72 頁
Now the boundary conditions at x = 0, and x = give
These leads to
c1 = 0,
That is,
Ch11-14 第 73 頁
電磁學例題
2017年1月24日 上午 11:42
Ch11-14 第 74 頁
補充:Oversized Fourier series與Gibb's phenomenon
2017年2月13日 下午 02:43
本文探討正交基底的一些缺失
Fourier series所用的基底為{1,cos(n*pi*x/p),sin(n*pi*x/p)}
週期T=2p=a的函數
現在,考慮加入T=2a的sin(pi*x/a)和cos(pi*x/a)兩項基底
若我們在傅立葉展開中加入這兩個基底
由half range expansion的概念,在x=a處
cos(pi*x/a)會有從-1跳到+1的點(jump),微分後會出現delta-
function
sin(pi*x/a)則會有折點,無論是jump還是折點
在物理上都代表有能量進入
在數學上,則代表邊界可有更多種的組合變化
說明如下:
考慮邊界是x=0和x=a
sin(pi*x/a)在x=0和x=a都是0,一次微分非0,二次微分又是0
=>邊界奇次微分不為0,偶次為0
cos(pi*x/a)在x=0是1,x=a是-1都不等於0,但一次微分後又是0
=>邊界偶次微分為0,奇次為0
如此無論在各種次數的微分值上,都可以由這兩個函數去
做線性組合表達
故這兩項基底在物理和數學上都可以complete(具有完備
性)
Ch11-14 第 75 頁
補充-多樣化邏輯與單一性邏輯
2017年2月13日 下午 06:34
1.單一性邏輯:如車子燃燒汽油,經過一連串sequential(串
列)的操作,排放廢氣驅動車輪前進,此為單一化邏輯。要
抓到出問題的環節很容易,只需要查看哪個環節堵塞出
錯。
2.多樣性邏輯:人體的運作是多樣化邏輯,是有
recirculate(再循環)的效果。例如人體的水會經由腎回收
再利用,就是多樣性邏輯的展現。但如此一來,要找出病
因就不是很容易,因為每個器官間的交互作用就會影響
到結果。
一般教科書解微分方程會像是是單一性邏輯的操作(特
定的SOP解法)
但本書從singular points來看,是用多樣化邏輯來探討
解的存在與多種的可能性
Ch11-14 第 76 頁
2017年2月17日 上午 11:04
Ch11-14 第 77 頁