Christine Kreuzer wrote:
harminv often returns negative Q values.
Yes, this is to be expected in a lossless, or nearly lossless system.
Remember that Q is proportional to 1 / (imaginary part of frequency).
If the imaginary part of the frequency is zero, numerically it will be
some tiny number and as a result Q will be some huge number with a
meaningless sign.
I have a few questions about harminv and the values returned by harminv.
If I got it right, in the thread about negative Q values you say that
the error bar returned by meep should be much smaller than the decay
rate to obtain a trustable and positive Q value.
Is the decay rate identic to the imag. freq. returned by harminv?
Yes. Or, at least within a factor of two (depending upon whether you are
talking about the decay rate of the field or the energy).
When I looked for the interpretation of the error given by harminv on
the harminv man page there is written that the error is a "crude
estimate of the relative error in the complex frequency" and that this
value should just be as small as possible. I didn't find any value for
the error in relation to the imag. freqency.
By error in the complex frequency, I mean the |(error in real part) + i
* (error in imaginary part)|. Harminv doesn't report separate error
estimates for the real and imaginary parts.
Can I improve the results of Q when I choose different fcen and df for
the source and harminv?
Yes. In general, harminv does a better job the narrower the source is
around the frequency of interest to you (since this filters out the
"noise" from all the other modes).
Steven
_______________________________________________
meep-discuss mailing list
[email protected]
http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/meep-discuss
