[Nek5000-users] fft in nek5000 or Visit ?

[nek5000-users at lists.mcs.anl.gov]

Dear Neks

 

I have my turbulent flow in a 3D box for which I would like to perform a
Fourier analysis in one direction (y) at several selected time steps in
order to find the discretized spectral representation (up to resolution of
say N=128 wave numbers) of streamwise vorticity (w_x). I have been trying to
find the best way to approach this in nek by reading the user mailing list
but am troubled to decide and would appreciate it if the experienced users
could guide me:

 

1)      Using int_tp tool (in repo, trunk/tool/) one can interpolate the nek
SEM output results to uniform, SEM and non-uniform (selected points) but I
guess this only works for small grids? At least I got seg fault for my
larger grids. Alternatively, we can use "ifreguo = .true." in userchk via
post-processing the data in parallel. And then use a separate fft library
(eg. MATLAB) to read in the uniform grid and do fft. But there will be
memory issues for larger datasets. How can I resolve this?

2)      I noticed that Visit has an FFT operator (at least in my v 2.6.3)
for which I could not find any documentation. Does anyone have any knowledge
of how it works?

3)      One could also do the fft by adding proper subroutines in .usr file
and calling them in userchk. Does anyone here have the right set of
subroutines for this purpose?

 

Thank you,

Hesam

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/nek5000-users/attachments/20140104/ec2bc41d/attachment.html>