Hi Ian,
I did something like the last idea several times. Instead of saving all
files I had a matlab program running in the background which loaded the
fields from hdf5, did the dft summing and then deleted the hdf5 file. In
this way only hardisk space for one field file is needed. When you have
a multicore machine this will work nicely. The only overhead you have is
the hdf5 file saving/loading. I did not save every time step, but only
enough to have the frequencys of interest properly sampled.
Andreas
Ian Buss wrote:
Hello all,
I would like to be able to define a function in my ctl file to declare a
plane across which the DFT is accumulated (and stored in an hdf5 file)
at every point at a single frequency. So for a 10x10 plane with
resolution 10, there would be 100x100xN complex data points in my hdf5
file, where N is the number of time steps (also to be specified via
(at-every) or something).
Examining the C++ interface, I see that field::add_dft is the sort of
thing I need. However, I am unsure of how to expose this using the SWIG
interface. I know I need to do something like (new-meep-dft-chunk...)
and then define this somehow with (meep-fields-add-dft c plane freq freq
1 false), where c is the component and plane is a 2D (volume) type. Am
I along the right lines here? How can I mix this with a (to-appended )
command to get the hdf5 file? Could someone help with the nuts and
bolts? Shall I abandon ship and use the C++ interface?
Of course, I could just declare something like:
(to-appended "p1" (at-every 0.075 (in-volume (volume (center 0 0 0)
(size 10 10 0)) output-efield-x)))
and then analyse using post processing, but more than a few of these
swiftly fills up a hard disk (when each file can be a few GB). I am
also aware that a CW source at the same frequency can provide
essentially the same information, but again this becomes tedious for
more than a few frequencies of interest. Any help would be gratefully
received,
Kind regards,
Ian Buss
---------------------------
Research PhD Student
Photonics Group
Dept. of Electrical & Electrical Engineering
University of Bristol
Rm 2.11, Queen's Building
University Walk
Bristol
BS8 1TR
email: [email protected]
_______________________________________________
meep-discuss mailing list
[email protected]
http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/meep-discuss
