Dear Steven and meep users,
first of all I'would like to thank you
for the recently arising activity in this
maillist.
In my last experiments with meep I've found
that I did not understand the flux computation
principles correctly. I have been trying to compute
the CW signal propagation through 2 dimensional
periodic structure. My main goal in this computational
task was to sort out nonlinearity like Kerr nonlinearity
in meep and to understand the flux computation in meep in
order to estimate the nonliner additive to the linear part
of the refraction index (effective change in refractive index).
I've understand that for this estimation I have to
compute the power that I get in my waveguide.
I've made this with (flux-in-box ...) function
in the ctl-file-interface.
I define the flux region to the flux computation
and I've defined a procedure, that compute flux in box,
i.e. in the specific space region:
;#####################################################
(define (flux-plane) (print "flux-poynting-plane:, "
(meep-time) ", " (flux-in-box X (volume (center x2 0 0) (size 0 sy 0)))
"\n"))
;#####################################################
And finally I've written this in the end of my ctl file:
;#####################################################
(run-until Tmax
(at-every dt flux-plane)
)
;#####################################################
and after grepping the output file I import into my
plotting program and see the time dependence of the flux.
I suppose, that it is the absolute value of S vector
(poynting vector) versus time in my picture.
I see the value of S: it is about 20, for example.
My Kerr susceptibility parameter
(define-param xx3 1e-2)
I choose the units of length in microns. Let the power
being in units of Watts (the units of n2 therefore is un^2/W).
The value of the product n2*power is then approximately
about 1e-1. Let it be so.
Afterwards, I've made the flux computation in this
way
;#####################################################
(define transx1 ; at the x1-point
(add-flux (* 0.5 (+ fmin fmax)) (- fmax fmin) nfreq
(make flux-region (center x2 0)
(size 0 (* 1 sy) 0)
)
)
)
(display-fluxes transx1)
;#####################################################
And I import into plotting program again and see the
frequency dependence of the flux. And on the amplitude axis
I see 100000 (that is aplitude of the frequency range near
the center frequency). And at this point I'm in a stupor...
That means that the nonlinear additive is greater that the
linear part of refractive index n ( (1e+5) * (1e-2) = 1e+3 ).
The amplitude of the input signal is about 10, frequency 0.3600.
There is no extreme value in the parameters I've used.
I apprehend danger in the fact of my incomprehension of
the flux core meaning in electrodynamics and meep simulation.
And very long ago I ask myself and friends in this maillist
why there are negative value in flux-spectrum, if it is fast
fourier transform? I interpret it for myself as in the
system I have considered exists 2 fluxes - one of them
propagate in the "+" direction and other in the "-" direction.
And then we do a fft with the former and the latter and
the next step is to subtract the second from the first - that
is because flux-spectrum in meep can have a negative values.
And I suppose that the flux spectrum is not the spectrum
of poynting vector value, because the poynting vector
spectrum has constant component and the peak at the
doubled frequency of the input CW source. But in the
flux-spectrum that I've obtained (as I've written above)
the peak was exactly near 0.3600. It corresponds the
CW source frequency.
Sorry for this long long letter)
Thank you very much for you help.
Best regards,
Sasha.
Undergraduated student of the Saratov State University, Russia.
_______________________________________________
meep-discuss mailing list
[email protected]
http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/meep-discuss
