当前位置:网站首页>Derivation of Fourier transform
Derivation of Fourier transform
2022-07-03 09:20:00 【fishfuck】
function f L ( t ) f_L(t) fL(t) With L L L ( L > 0 ) (L>0) (L>0) For cycles , In the interval [ − L 2 , L 2 ] [-\frac L2,\frac L2] [−2L,2L] Continuous on ( Or meet the Dirichlet condition ), be f L ( t ) f_L(t) fL(t) stay [ − L 2 , L 2 ] [-\frac L2,\frac L2] [−2L,2L] Can be expanded into Fourier series
f L ( t ) = a 0 2 + ∑ n = 1 ∞ ( a n cos n ω t + b n sin n ω t ) . f_{L}(t)=\frac{a_{0}}{2}+\sum_{n=1}^{\infty}\left(a_{n} \cos n \omega t+b_{n} \sin n \omega t\right) \text {. } fL(t)=2a0+n=1∑∞(ancosnωt+bnsinnωt).
among
ω = 2 π L a n = 2 L ∫ − L 2 L 2 f L ( t ) cos n ω t d t . n = 0 , 1 , 2 … b n = 2 L ∫ − L 2 L 2 f L ( t ) sin n ω t d t . n = 1 , 2 , ⋯ \begin{aligned} &\omega=\frac{2\pi}{L} \\ &a_{n}=\frac{2}{L} \int_{-\frac{L}{2}}^{\frac{L}{2}} f_{L}(t) \cos n \omega t d t . \quad n=0,1,2 \ldots \\ &b_{n}=\frac{2}{L} \int_{-\frac{L}{2}}^{\frac{L}{2}} f_{L}(t) \sin { n \omega t } d t . \quad n=1,2, \cdots \end{aligned} ω=L2πan=L2∫−2L2LfL(t)cosnωtdt.n=0,1,2…bn=L2∫−2L2LfL(t)sinnωtdt.n=1,2,⋯
So there are
cos t = e i t + e − i t 2 \cos t=\frac{e^{i t}+e^{-i t}}{2} cost=2eit+e−it
sin t = − i e i t − e − i t 2 \sin t=-i \frac{e^{i t}-e^{-i t}}{2} sint=−i2eit−e−it
be
f L ( t ) = a 0 2 + ∑ n = 1 ∞ ( a n cos n ω t + b n sin n ω t ) = a 0 2 + ∑ n = 1 ∞ ( a n e i n ω t + e − i n ω t 2 − i b n e i n ω t − e − i n ω t 2 ) = a 0 2 + ∑ n = 1 ∞ ( a n − i b n 2 e i n ω t + a n + i b n 2 i e − i n ω t ) \begin{aligned} f_{L}(t) &=\frac{a_{0}}{2}+\sum_{n=1}^{\infty}\left(a_{n} \cos n \omega t+b_{n} \sin n \omega t\right) \\ &=\frac{a_{0}}{2}+\sum_{n=1}^{\infty}\left(a_{n} \frac{e^{i n \omega t}+e^{-i n \omega t}}{2}-i b_{n} \frac{e^{i n \omega t}-e^{-i n \omega t}}{2}\right) \\ &=\frac{a_{0}}{2}+\sum_{n=1}^{\infty}\left(\frac{a_{n}-i b_{n}}{2} e^{i n \omega t}+\frac{a_{n}+i b_{n}}{2 i} e^{-i n \omega t}\right) \end{aligned} fL(t)=2a0+n=1∑∞(ancosnωt+bnsinnωt)=2a0+n=1∑∞(an2einωt+e−inωt−ibn2einωt−e−inωt)=2a0+n=1∑∞(2an−ibneinωt+2ian+ibne−inωt)
We make
c 0 = a 0 2 = 1 L ∫ − ∞ ∞ f L ( t ) d t c_{0}=\frac{a_{0}}{2}=\frac{1}{L} \int_{-\infty}^{\infty} f_{L}(t) d t c0=2a0=L1∫−∞∞fL(t)dt
c n = a n − i b n 2 c_{n}=\frac{a_{n}-i b_{n}}{2} cn=2an−ibn
c − n = a n + i b n 2 {c_{-n}}=\frac{a_{n}+i b_{n}}{2} c−n=2an+ibn
So there are
C n = 1 L ∫ − L 2 L 2 f L ( t ) e − i n ω t t d t n = 0 , ± 1 , ± 2 , … C_{n}=\frac{1}{L} \int_{-\frac{L}{2}}^{\frac{L}{2}} f_{L}(t) e^{-i n \omega t t} d t \quad n=0, \pm 1,\pm 2, \ldots Cn=L1∫−2L2LfL(t)e−inωttdtn=0,±1,±2,…
be
f L ( t ) = c 0 + ∑ n = 1 ∞ ( c n e i n ω t + c i n e − i n ω t ) = ∑ n = − ∞ ∞ C n e i n u t t = ∑ n = − ∞ ∞ ( 1 2 ∫ − L 2 L 2 f L ( τ ) e − i n ω t d τ ) e i n ω t \begin{aligned} f_{L}(t) &=c_{0}+\sum_{n=1}^{\infty}\left(c_{n} e^{i n \omega t}+c_{i n} e^{-i n \omega t}\right) \\ &=\sum_{n=-\infty}^{\infty} C_{n} e^{i n u t t} \\ &=\sum_{n=-\infty}^{\infty}\left(\frac{1}{2} \int_{-\frac{L}{2}}^{\frac{L}{2}} f_{L}(\tau) e^{-i n \omega t} d \tau \right) e^{i n \omega t} \end{aligned} fL(t)=c0+n=1∑∞(cneinωt+cine−inωt)=n=−∞∑∞Cneinutt=n=−∞∑∞(21∫−2L2LfL(τ)e−inωtdτ)einωt
We consider nonperiodic functions in the interval [ − L 2 , L 2 ] [-\frac L2,\frac L2] [−2L,2L] On the situation . Make
f ( t ) = ∑ n = − ∞ ∞ c n e i n ω t ( − L 2 < t < L 2 ) . ∗ f(t)=\sum_{n=-\infty}^{\infty} c_{n} e^{ {in\omega t }}\left(-\frac{L}{2}<t<\frac{L}{2}\right). \quad* f(t)=n=−∞∑∞cneinωt(−2L<t<2L).∗
among
c n = 1 L ∫ − L 2 L 2 f ( t ) e − i n ω τ d τ . c_{n}=\frac{1}{L} \int_{-\frac{L}{2}}^{\frac{L}{2}} f(t) e^{-i n \omega \tau} d \tau . cn=L1∫−2L2Lf(t)e−inωτdτ.
We noticed that ( ∗ ) (*) (∗) The right end of the equation defines a period of L L L Function of f L 1 ( t ) = { f ( t ) − L 2 < t < L 2 f ( − L 2 ) + f ( L 2 ) 2 t = ± L 2 f_{L_{1}}(t)=\left\{\begin{array}{l}f(t) \quad-\frac{L}{2}<t<\frac{L}{2} \\ \frac{\left.f\left(-\frac{L }{2}\right)+f(\frac{L}{2}\right)}{2} \quad t=\pm \frac{L}{2}\end{array}\right. fL1(t)={ f(t)−2L<t<2L2f(−2L)+f(2L)t=±2L
Make L → + ∞ L \rightarrow+\infty L→+∞ Then periodic function f L 1 ( t ) f_{L_{1}}(t) fL1(t) The limit is f ( t ) f(t) f(t), That is to say
lim L → + ∞ f L 1 ( t ) = f ( t ) \lim _{L \rightarrow+\infty} f_{L_{1}}(t)=f(t) L→+∞limfL1(t)=f(t)
here , To any − ∞ < t < + ∞ -\infty<t<+\infty −∞<t<+∞, Yes
f ( t ) = lim L → + ∞ ∑ n = − ∞ + ∞ c n e i n ω t . f(t)=\lim _{L \rightarrow+\infty} \sum_{n=-\infty}^{+\infty} c_{n} e^{ {in\omega t. }} f(t)=L→+∞limn=−∞∑+∞cneinωt.
( ∗ ) (*) (∗) The right side of the formula is ∑ n = − ∞ + ∞ c n e i n ω t . \sum_{n=-\infty}^{+\infty} c_{n} e^{ {in\omega t. }} ∑n=−∞+∞cneinωt., among ω = 2 π L \omega=\frac{2 \pi}{L} ω=L2π.
Make
g n = c n L = ∫ − L 2 L 2 f ( t ) e − i n 2 π t L d t g_{n}=c_{n} L=\int_{-\frac{L}{2}}^{\frac{L}{2}} f(t) e^{\frac{-in 2 \pi t} L} d t gn=cnL=∫−2L2Lf(t)eL−in2πtdt
be
f L ( t ) = 1 2 π ∑ n = − ∞ ∞ g n e − i n 2 π t L [ ( n + 1 ) − n ] 2 π L = ∑ n = − ∞ ∞ ( 1 L ∫ − L 2 L 2 f L ( τ ) e − i n ω τ d τ ) e i n ω t \begin{aligned} f_{L}(t) &=\frac{1}{2\pi} \sum_{n=-\infty}^{\infty} g_{n} e^{-\frac{i n 2 \pi t}{L}} \frac{[(n+1)-n] 2\pi}{L} \\ &=\sum_{n=-\infty}^{\infty}\left(\frac{1}{L} \int_{-\frac{L}{2}}^{\frac{L}{2}} f_{L}(\tau) e^{-i n \omega \tau} d \tau\right) e^{ {in\omega t }} \end{aligned} fL(t)=2π1n=−∞∑∞gne−Lin2πtL[(n+1)−n]2π=n=−∞∑∞(L1∫−2L2LfL(τ)e−inωτdτ)einωt
remember ω n = n 2 π L \omega_{n}=\frac{n 2 \pi}{L} ωn=Ln2π, We make F [ f ( t ) ] L ( u ) = ∫ − L 2 L 2 f ( t ) e − i ω t d t \mathscr{F}\left[f(t) \right]_{L}(u)=\int_{-\frac{L}{2}}^{\frac{L}{2}} f(t) e^{-i \omega t} d t F[f(t)]L(u)=∫−2L2Lf(t)e−iωtdt, be :
f L ( t ) = 1 2 π ∑ n = − ∞ ∞ F [ f ( t ) ] L ( ω n ) e i ω n t ( ω n + 1 − ω n ) . ∗ ∗ f_{L}(t)=\frac{1}{2 \pi} \sum_{n=-\infty}^{\infty} \mathscr{F}[f(t)]_{L}\left(\omega_{n}\right) e^{ {i\omega n } t}\left(\omega_{n+1}-\omega_{n}\right) \text {. } \quad** fL(t)=2π1n=−∞∑∞F[f(t)]L(ωn)eiωnt(ωn+1−ωn). ∗∗
When L → ∞ L \rightarrow \infty L→∞ when , F [ f ( t ) ] L ( ω ) \mathscr{F}[f(t)]_{L}\left(\omega\right) F[f(t)]L(ω) Naturally, it tends to a frequency domain function F [ f ( t ) ] ( ω ) = ∫ − ∞ ∞ f ( t ) e − i ω t d t \mathscr{F}[f(t)](\omega)=\int_{-\infty}^{\infty} f(t) e^{-i \omega t} d t F[f(t)](ω)=∫−∞∞f(t)e−iωtdt. So we finished the time domain function f ( t ) f(t) f(t) To frequency domain function F [ f ( t ) ] ( ω ) \mathscr{F}[f(t)]\left(\omega\right) F[f(t)](ω) Transformation of . We call F [ f ( t ) ] ( ω ) \mathscr{F}[f(t)]\left(\omega\right) F[f(t)](ω) It's Fourier transform .
With L → ∞ L \rightarrow \infty L→∞, Δ ω n = w n + 1 − w n → 0 \Delta \omega_{n}=w_{n+1}-w_{n} \rightarrow 0 Δωn=wn+1−wn→0, be ( ∗ ∗ ) (**) (∗∗) The formula can be regarded as 1 2 π ∫ − ∞ + ∞ F [ f ( t ) ] ( ω ) e i ω t d ω \frac{1}{2\pi} \int_{-\infty}^{+\infty} \mathscr{F}[f(t)](\omega) e^{i \omega t} d\omega 2π1∫−∞+∞F[f(t)](ω)eiωtdω Riemann sum of , So export
f ( t ) = 1 2 π ∫ − ∞ ∞ F [ f ( t ) ] ( ω ) e i ω t d ω . f(t)=\frac{1}{2\pi} \int_{-\infty}^{\infty} \mathscr{F}[f(t)](\omega) e^{i \omega t} d \omega . f(t)=2π1∫−∞∞F[f(t)](ω)eiωtdω.
Call the above formula F [ f ( t ) ] ( ω ) \mathscr{F}[f(t)](\omega) F[f(t)](ω) Inverse Fourier transform .
To what is known g ( ω ) g(\omega) g(ω), Its inverse Fourier transform is recorded as
F − 1 [ g ( ω ) ] ( t ) = 1 2 π ∫ − ∞ ∞ g ( ω ) e i ω t d ω \mathscr{F}^{-1} [g(\omega)](t)=\frac{1}{2 \pi} \int_{-\infty}^{\infty} g(\omega) e^{i \omega t} d \omega F−1[g(ω)](t)=2π1∫−∞∞g(ω)eiωtdω
边栏推荐
- Use the interface colmap interface of openmvs to generate the pose file required by openmvs mvs
- Linxu learning (4) -- Yum and apt commands
- LeetCode 324. Swing sort II
- What are the stages of traditional enterprise digital transformation?
- 精彩回顾|I/O Extended 2022 活动干货分享
- Save the drama shortage, programmers' favorite high-score American drama TOP10
- [point cloud processing paper crazy reading frontier edition 13] - gapnet: graph attention based point neural network for exploring local feature
- IDEA 中使用 Hudi
- AcWing 786. 第k个数
- Principles of computer composition - cache, connection mapping, learning experience
猜你喜欢

LeetCode 1089. 复写零

Just graduate student reading thesis

【点云处理之论文狂读前沿版10】—— MVTN: Multi-View Transformation Network for 3D Shape Recognition

LeetCode 508. 出现次数最多的子树元素和

State compression DP acwing 291 Mondrian's dream

一个优秀速开发框架是什么样的?
![[point cloud processing paper crazy reading frontier version 8] - pointview gcn: 3D shape classification with multi view point clouds](/img/ee/3286e76797a75c0f999c728fd2b555.png)
[point cloud processing paper crazy reading frontier version 8] - pointview gcn: 3D shape classification with multi view point clouds

Redis learning (I)

传统企业数字化转型需要经过哪几个阶段?

In the digital transformation, what problems will occur in enterprise equipment management? Jnpf may be the "optimal solution"
随机推荐
拯救剧荒,程序员最爱看的高分美剧TOP10
[point cloud processing paper crazy reading frontier version 8] - pointview gcn: 3D shape classification with multi view point clouds
MySQL installation and configuration (command line version)
图像修复方法研究综述----论文笔记
【点云处理之论文狂读经典版11】—— Mining Point Cloud Local Structures by Kernel Correlation and Graph Pooling
Bert install no package metadata was found for the 'sacraments' distribution
Sword finger offer II 091 Paint the house
CSDN markdown editor help document
Just graduate student reading thesis
What is the difference between sudo apt install and sudo apt -get install?
Matlab dichotomy to find the optimal solution
Digital statistics DP acwing 338 Counting problem
【点云处理之论文狂读前沿版10】—— MVTN: Multi-View Transformation Network for 3D Shape Recognition
STM32F103 can learning record
In the digital transformation, what problems will occur in enterprise equipment management? Jnpf may be the "optimal solution"
Build a solo blog from scratch
2022-2-13 learning the imitation Niuke project - home page of the development community
Instant messaging IM is the countercurrent of the progress of the times? See what jnpf says
What is an excellent fast development framework like?
Methods of using arrays as function parameters in shell