fyi-
this is indeed what was happening. if i reduce df (frequency width of
the gaussian pulse), the oscillations do go away OR if i simply run
the simulation much longer (e.g. time ~ 400 * df) than what energy
decay implies the oscillations also vanish.
too cool and thanks!
gp
------
On Aug 22, 2006, at 5:38 PM, Steven G. Johnson wrote:
On Thu, 17 Aug 2006, G.J. Parker wrote:
what i'm seeing is that R and T are NOT the same at each plane. in
particular, at large frequencies, there is some spurious (?)
oscillations in R & T which depend on the position of the flux
planes. since there is nothing but air between the various flux
planes, it appears energy is not conserved.
A couple possibilities come to mind, both related to the fact that
you are using a very sharp input pulse to get a very broad
bandwidth at once.
First, I seem to recall that the Fourier transforms run into some
kind of numerical precision problems if you try to get too broad a
bandwidth at once. (This is something I never looked at in detail,
just something I remember from our old Fortran code years
ago...since then I rarely use bandwidths more than 50%, just from
habit.)
Second, if you have a sharp feature in the spectrum around there,
you may not be running long enough for the spectrum to be converged
thanks to the usual uncertainty principle. Remember, you have a
short broad-band pulse, so only a tiny portion of your incident
energy may be at the resonant frequency---this makes looking at the
decay of the total energy, as you are doing, possibly deceptive.
Steven
_______________________________________________
meep-discuss mailing list
[email protected]
http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/meep-discuss
_______________________________________________
meep-discuss mailing list
[email protected]
http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/meep-discuss
