I agree with Gursoy, your resolution appears to be too low. His rule of thumb of a minimum of 8 (I actually prefer 10 - 20) pixels or grid points for one wavelength is good for wave propagation in lossless dielectrics. Lossy materials, like real metals typically need much higher resolution. My rule of thumb is that you need a minimum of 1 or 2 grid points per skin depth of your lossy material. You can calculate skin depth from the n and k (real and imaginary parts of refractive index) of silver around your wavelength of interest.
Another piece of advice that has been expressed many times in this mailing list is to run several simulations each with increased resolution until you get two successive results that "do not change much." Good luck, Neal On Thu, May 30, 2013 at 2:23 AM, gursoy akguc <akgu...@yahoo.com> wrote: > I think your resolution is way low for this system. as far as I can see > your geometry is 700nm cube > res 32 means your pixel cell size is about 22nm. source is 363nm and in > the dielectric material you use 8 it means wavelength is about 45nm. rule > of thumb is 1 wavelength needs to be represented by 8 pixels, but your 8 > pixels makes 176nm about 4 times higher than wavelength. > my suggestion is to use 120resolution minimum to see some meaningful > result comparable to Mie. > > Gursoy B. Akguc > > *From:* niels jelsma <c.j.m.jel...@student.rug.nl> > *To:* meep-discuss@ab-initio.mit.edu > *Sent:* Wednesday, May 29, 2013 2:23 PM > *Subject:* [Meep-discuss] Resolution dependence silver nanosphere > > Dear All, > > I'm trying to calculate the dipolemoment of a silver nanosphere in Meep. > In the end I am interested in looking at a chain of spheres and the > transmittance, p_last / p_first. > > However for just a single sphere I don't understand what values I have > to choose for the resolution to get proper results. > I've read on the mailing list that I should choose a resolution that > gives at least 8pixels/wavelength and then double the resolution to get > better results. > However, increasing the resolution this way sometimes gives me worse > results. > > A simplified version of my script can be found here: > https://gist.github.com/maxnuf/5669273 > > I've setup a sphere with radius 25nm in 3d space and a continuous source > as plane-wave. > After running it for a while to get into a steady-state I integrate the > Polarization in a volume containing the sphere to find the dipolemoment. > > I expect to find a peak near the resonant frequency. So I run the > simulation for several input source wavelengths. > > However, for different resolutions, I get completely different results. > I don't know what to do to remedy my problem. > > As can be seen in this figure: > http://postimg.org/image/kua54kwq9/ > > At a resolution 32 I see a sort of dip where I expect the resonant > frequency (which should be at 363nm according to Mie theory) > > I'm using Meep 1.2 and a Drude-model for my silver sphere. > > How can I determine which resolutions are suitable? > > Thanks, > Niels > > > _______________________________________________ > meep-discuss mailing list > meep-discuss@ab-initio.mit.edu > http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/meep-discuss > > > > _______________________________________________ > meep-discuss mailing list > meep-discuss@ab-initio.mit.edu > http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/meep-discuss >
_______________________________________________ meep-discuss mailing list meep-discuss@ab-initio.mit.edu http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/meep-discuss
