v*********r 发帖数: 8 | 1 Given a cubic domain with length L in each direction. The size of the grid in
this domain is say, 128 cube. Velocity components at all grid points are
known. How to calculate the kinetic energy spectrum E(k)?
The primal confusion is how to transform the 3D space velocity field into 3D
Fourier space field. Hope to get detailed explanation...
Thanks a lot for your patience!!! |
sc 发帖数: 122 | 2 x,y,z上速度分开做福利也变换
【在 v*********r 的大作中提到】![](/moin_static193/solenoid/img/up.png) : Given a cubic domain with length L in each direction. The size of the grid in : this domain is say, 128 cube. Velocity components at all grid points are : known. How to calculate the kinetic energy spectrum E(k)? : The primal confusion is how to transform the 3D space velocity field into 3D : Fourier space field. Hope to get detailed explanation... : Thanks a lot for your patience!!!
|
v*********r 发帖数: 8 | 3 Thanks a lot!
What I think is that I have to do 128^2 1D fourier transf in each direction.
So for each velocity component I will do 3*128^2 fourier transf to get its
spectrum in(k1,k2,k3) space. Then calculate energy in (k1,k2,k3) space.
after this, sum energy in each wave number shell.
Is that right?
grid in
3D
【在 sc 的大作中提到】![](/moin_static193/solenoid/img/up.png) : x,y,z上速度分开做福利也变换
|
sc 发帖数: 122 | 4 每个方向上还是128^3个数据吧?怎么变成128^2了?
【在 v*********r 的大作中提到】![](/moin_static193/solenoid/img/up.png) : Thanks a lot! : What I think is that I have to do 128^2 1D fourier transf in each direction. : So for each velocity component I will do 3*128^2 fourier transf to get its : spectrum in(k1,k2,k3) space. Then calculate energy in (k1,k2,k3) space. : after this, sum energy in each wave number shell. : Is that right? : : grid in : 3D
|
v*********r 发帖数: 8 | 5 128^2 1D fourier transf. not data points
direction.
【在 sc 的大作中提到】![](/moin_static193/solenoid/img/up.png) : 每个方向上还是128^3个数据吧?怎么变成128^2了?
|
sc 发帖数: 122 | 6 没看明白,难道不是128^3一块儿做了变换去?
【在 v*********r 的大作中提到】![](/moin_static193/solenoid/img/up.png) : 128^2 1D fourier transf. not data points : : direction.
|
v*********r 发帖数: 8 | 7 there are 128^2 line in x direction, right.
each line has 128 data points.
u need to 1d fourier tranf on each line over the 128 points respectively.
so there should be 128^2 tranf in x direction.
Then u will do y and z direction subsequently.
【在 sc 的大作中提到】![](/moin_static193/solenoid/img/up.png) : 没看明白,难道不是128^3一块儿做了变换去?
|
sc 发帖数: 122 | 8 你这个fft做的奇怪,随便搞一个fftw直接做3d的不就完了?
【在 v*********r 的大作中提到】![](/moin_static193/solenoid/img/up.png) : there are 128^2 line in x direction, right. : each line has 128 data points. : u need to 1d fourier tranf on each line over the 128 points respectively. : so there should be 128^2 tranf in x direction. : Then u will do y and z direction subsequently.
|
v*********r 发帖数: 8 | 9 I think in theory should be like what I said.
But what is fftw? sounds cool.
thanks for sharing
【在 sc 的大作中提到】![](/moin_static193/solenoid/img/up.png) : 你这个fft做的奇怪,随便搞一个fftw直接做3d的不就完了?
|