How is an autocorrelation function computed? (Dynamic Light Scattering)

1 reply · 2K views · last reply
Salmone
Messages
101
Reaction score
13
In an experiment of Dynamic Light Scattering, how is an autocorrelation like the one in the image computed?

Semi-log plot of intensity autocorrelation against time in microseconds, from 0.1 to about 1000000 us. The curve starts near 1.2, falls steeply between roughly 10 and 1000 us, and levels off at 1.

Mathematically a correlation function can be written as ##G(\tau)=\langle I(t)I(t+\tau) \rangle##, in an experiment like the one I mentioned the scattered intensity light is collected by a single detector, then the signal is sent to a digital correlator which computes the correlation function. How this process works? Once I have a signal from the detector, what does the correlator do? Does it multiply the intensity at time ##t## with the same intensity at time ##t+\tau## simply? How is the average implemented? By recording with the detector the same scattered intensity multiple times? Can you explain very generally how a digital correlator works?
 
Salmone said:
Can you explain very generally how a digital correlator works?
A correlator would take the FFT of the two signals, multiply those two spectra, then inverse FFT. In effect, filtering a signal by another signal.

I expect autocorrelation could be performed by taking the FFT of the signal, squaring the vectors of the resulting spectrum, then computing the inverse FFT. In effect, filtering a signal by itself.