On May 7, 2008, at 12:41 PM, sq wrote: > Dear Steven and Meep users, > I want to monitor the time evolution of electric field at a fixed > point in > vacuum. But I can not understand the results from the simulation. > Firstly, I set two monitors in the simulation. One is close to the > source, and > the other is far from the source. However, the amplitude of two > electric > fields is different each other. The amplitude of the electric field > closed to > the source is larger than that of far from the source. In my > opinion, the > amplitude of two electric fields should be equal as propagating in > vacuum.
That's certainly not true in vacuum, because the fields of a point source decay as 1/r in 3d and as 1/sqrt(r) in 2d. (This is required simply by conservation of energy, so that the power flux is independent of r.) Even if you have a waveguide or some other 1d-like system where the fields are not decaying as they propagate, there can still be e.g. radiating evanescent fields that make the field larger closer to the source. > Meanwhile, the amplitude of two electric fields is far less than 1. Why do you expect it to be 1? When you specify the amplitude of the source in Meep, you are specifying the amplitude of the *current*. This does not have any simple relationship to the amplitude of the resulting field, because the relationship depends on geometry. The same current source in two different geometries will emit different field amplitudes and different amounts of power. > Secondly, when I zoom in the two sides of the electric field (at the > beginning > and the end of time), I find that there is some noise. If I change the > Gaussian source into continuous source, the result is worse. I think > that the > two sides should be smooth to get a perfect simulated result. What > is the > reason? How can I improve it? Figure 2 and figure 3 give the results. > I attached the results in the previous email, but the figures are > too big to > post them. Attachment is my ctl file. Thanks a lot. There are probably two things that you are seeing. First, when you turn on the source (and even for the Gaussian, depending on the "cutoff" parameter, there is some sharp turn-on in the tail of the Gaussian), you are going to get high-frequency transient behaviors because of the sharp turn on. Second, if you have e.g. a Gaussian source and you run for a long time and wait until the fields have decayed away, what you have left over is a combination of numerical noise and very high-frequency waves (near the Nyquist frequency) that have low group velocity due to numerical dispersion. Regards, Steven G. Johnson _______________________________________________ meep-discuss mailing list [email protected] http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/meep-discuss
