Questions about FDFD
时间：03-29
整理：3721RD
点击：
I've written a FDFD simulation in python following the procedure written by Dr. Rumpf here: http://emlab.utep.edu/ee5390cem/Lect...ementation.pdf
I think I'm having some issues with the source.
Slide 36 defines k0 as 2*pi / wavelength
If I let wavelength be 1000e-9 meters or 1000 nm k0 = 6283185.30718
nref = 1
kinc = k0 * nref * [0,1]
kinc then should be a source from the y direction
slide 40 defines the source field to be
fsrc (x,y) = exp(j *(kxinc X + kyinc Y) )
fsrc (x,y) = exp(j *( 0 + 6283185.30718 * Y) )
or f = exp(j * 6283185.30718 * n) ) n = 0..M
which just becomes a decaying function (because we aren't resolving a wavelength) unless I multiply my n term by some deltaX (1000e-10). Am i missing something somewhere?
I think I'm having some issues with the source.
Slide 36 defines k0 as 2*pi / wavelength
If I let wavelength be 1000e-9 meters or 1000 nm k0 = 6283185.30718
nref = 1
kinc = k0 * nref * [0,1]
kinc then should be a source from the y direction
slide 40 defines the source field to be
fsrc (x,y) = exp(j *(kxinc X + kyinc Y) )
fsrc (x,y) = exp(j *( 0 + 6283185.30718 * Y) )
or f = exp(j * 6283185.30718 * n) ) n = 0..M
which just becomes a decaying function (because we aren't resolving a wavelength) unless I multiply my n term by some deltaX (1000e-10). Am i missing something somewhere?
The 'j' in your exponential term makes it oscillate and not decay. Also, when you multiple your very large k by a very small Y, you will get a reasonable number in your exponential. I recommend normalizing everything. Normalize your dimensions by multiplying by k0. Normalize your k vector components by dividing by k0.
BTW, try asking questions through the "CEM Lectures" channel on YouTube. I do not check EDABoard that often, but YouTube notifies me right away when a question is asked.
https://www.youtube.com/channel/UCPC...SK71MnPPcp8AGA
Happy to help you!