Hi Dr. Johnson and other MEEP knowledgeables!
At long simulation times I observe an interesting effect of a non-decaying
oscillatory mode within the calulated volume. The magnitude of the
oscillations remains at the level of 1e-4 - 1e-7 of the initial wave
magnitude.
Under some conditions the magnitude of oscillations start increasing
exponentially, thus making FDTD simulation impossible at long times.
This effect does not depend on the value of the Courant factor
and subpixel smoothing turning on/off.
I cite as an example simple ctl file specifying normal incidence
on infinite array of cylinders.
First two thousands iterations the field decreases to the level of 1e-3
of the initial wave magnitude and then it exponentially increases. After
10 thousands iterations the field value is of order of initial wave.
Do you have some explanation of this effect?
Could you recommend something in order to avoid it?
(set! geometry-lattice (make lattice (size 4 80 60)))
(set-param! resolution 1)
(set-param! Courant 0.5)
;(set-param! eps-averaging? false)
(set-param! k-point (vector3 0))
(set-param! ensure-periodicity #t)
(use-output-directory)
(set! geometry
(list
(make cylinder
(center 0 0 0)
(radius 5)
(height 4)
(axis (vector3 1 0 0))
(material
(make dielectric
(epsilon 4.41)
)
)
)
)
)
(define (berenger t) (* (- t 25) (exp (- 0 (sqr (/ (- t 25) 5))))))
(set! sources (list
(make source
(src (make custom-src (src-func berenger) (start-time 0) (end-time 100)))
(component Ex)
(center 0 0 -15)
(size 4 80 0)
)
)
)
(set! pml-layers (list (make pml (direction Z) (thickness 10))))
(run-until 10000
(at-beginning output-epsilon)
(at-every 500
(in-volume (volume (center 0 0 0) (size 0 80 60))
output-efield-x)
)
)
_______________________________________________
meep-discuss mailing list
[email protected]
http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/meep-discuss
