Dear MEEP-users,
for several weeks now we are trying to implement a near- to farfield
transformation
in MEEP, which will hopefully be a available for everyone soon. We
changed the MEEP-
source code to give the abillity to output the DFTs on a closed
rectangular surface
without averaging any fields and additionally output the positions of
the measurement
points. This step works fine, the DFT-fields are in agreement with
simulations done
in Lumerical.
In the second step, a python script was developed which does the NTFF as
discribed
in Taflove's book (3rd edition, 2005) plus geometric averaging of the
magnetic fields.
To test the procedure, a dipole antenna was simulated using a chain of
gaussian sources
of differing amplitudes, which produces very accurate nearfields.
The problem is now, that the NTFF only gives accurate results if the
monitors are
extremely close to the scatterer (antenna in this case). If the monitors
are even
a few Yee-cells away, the results turn out to be completely wrong! But
this behavior
has to be a bug, since there is no strong dependency of the monitor
positions
(except from numerical dispersion, which does not explain the dramatical
deviation
and which is even corrected to first order in our code).
Every part of the code is checked hundreds of times and we cannot find
any mistake.
Because of that, the question arose if the equations that we use, which
are from
Taflove, have to be changed because of the dimensionless units. Or if
anyone has
any idea/experience were the mistake could lie. In fact, it turned out
that the
impedance of free space, which is part of the equations, has to be used
with its
normal value of about 120pi, even though it is defined as
sqrt(mu0/epsilon0), which
should be 1 in MEEP-units. Any help is strongly appreciated!
Best regards,
Carlo Barth
_______________________________________________
meep-discuss mailing list
meep-discuss@ab-initio.mit.edu
http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/meep-discuss
