On Jul 21, 2011, at 4:00 PM, mf wrote:
The decay coefficient within any PML is well-known to have a
cos(theta) factor, where theta is the incidence angle, which makes
the decay become slower as you approach glancing incidence.
However, this is true even at infinite resolution, so if you are
seeing an effect that decreases with increasing resolution then it
is something different. e.g. it may be transition reflections,
which you can also decrease by making the PML thicker instead of
increasing the resolution.
Transition reflections also increase as you approach glancing
incidence, because at glancing incidence the phase-velocity
mismatch between incident and reflected waves goes to zero. See
e.g. equation (13) in our paper: http://www.opticsinfobase.org/abstract.cfm?URI=oe-16-15-11376
I realize this, and that is what I was afraid of. However the effect
seems rather strong: at a resolution of 64, the amplitude is .02 and
I can not run resolutions higher than 128 (where the oscillations
are still clearly visible). Is there anything I can do ? (modify the
functional absorption in the PML ?) My guess is that what I see are
reflections from the first PML layers (the results seem to do not
depend on the PML thickness).
If it is transition reflections, I don't see how they could fail to
depend on PML thickness. As you make the PML thicker and thicker, the
PML turns on more and more gradually (the same quadratic profile, but
shallower and more stretched out). Asymptotically for large
thicknesses, the transition reflections must vanish. The only caveat
is that the decrease with thickness does not occur until the PML is
sufficiently thick to be in the asymptotic regime -- normally, this is
true whenever the PML reflections are low, but maybe at your glancing
angles the reflection is currently so high that you are not yet in the
regime where you can see the thickness dependence. (See e.g. our
paper, mentioned previously, for some graphs.)
Try increasing the PML thickness by a factor of 10 to see if that
improves things. If it doesn't, I would begin to suspect that
something else is going on, although I'm not sure what it could be.
_______________________________________________
meep-discuss mailing list
[email protected]
http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/meep-discuss
