|Subject:||RE: phase binning light curves|
|Date:||Fri, 17 Jun 2011 10:03:26 +0800|
It sounds like what you want to do is cross correlation, which can give you an estimate of the phase difference between one curve, and another that is similar. This could be used to align your light curves with each other so that you can average “like” brightnesses. What it does is it takes your two curves and slides one over the other, bit by bit, finding a correlation coefficient each time. This is maximised when the one curve has been slid the correct phase difference over the other. So to find this you need to apply cross correlation, then look for the index of the maximum in the output of the function – that will tell you the phase difference.
It is in Octave as the xcorr function.
It is often categorised as a signal processing function more than a statistical function.
I'm trying to phase bin periodic light curves so that a periodically varying light curve over a period of days is folded onto itself and represented on a plot with a phased time axis of 0 to 1. I have a time and a brightness column and the method I'm using to process the data is:
1) take all of the values that lie at similar brightnesses and average them only with each other
2) for each different brightness average, i record where it first appeared on the plot it against the phased time value
This will ultimately give me a plot with one average curve representing all of the periods in the light curve.
Are there any statistical functions available in Octave that approximate this?
I've been searching but want to make sure that I don't miss any I may not have seen.
Just trying to get a general sense on how to proceed with this project.
|[Prev in Thread]||Current Thread||[Next in Thread]|