Example: tracker

0

This program produces an animated display which tracks the amplitude and phase of selected line features in the spectrum. It has a number of user-settable options which determine how the line is tracked. To run this program, type
tracker | xmgr -pipe
and an animated display will start up. In normal use, the parameters should be set as follows:

num_points: a power of two. A single phase/amplitude point is printed for each set of num_points samples.

padding_factor: a power of two. This determines the amount of padding done on the data set, and thus the ultimate frequency resolution of the line discrimination.

fpreset: your best guess for the frequency that you want to track. If the actual frequency of the spectral line differs from this value, then the phase will slowly drift as a linear function of time. (The tracker program does a robust best linear fit to this slope, and uses it to report a best frequency estimate.)

estimate: if set to zero, then the phase of the line found is always compared with the frequency preset above. If set non-zero, then tracker will make a ``best estimate" of the true frequency, and compare the phase of the line found with the phase appropriate to that sinusoid.

nbins: the number of (padded) frequency bins adjacent to the one of interest in which the line will be searched for. The frequency range covered is thus given by

$$\Delta f = \pm {{\tt nbins} \over \Delta t ({\tt num\_points} \times {\tt padding\_factor}+2)}$$ (16.29.334)

num_display: the number of points displayed by tracker. The total amount of time covered by the output is num_display $\times$ num_points where $\Delta t$ is the sample interval.

num_win,nwdt: these parameters are described in the section on multi-taper methods.

maxpass: the number of passes to make within remove_spectral_lines(). This number should be set as small as possible, provided that you still ``catch" the line of interest.

Figure: Output of the example program tracker, making use of remove_spectral_lines() to track the amplitude and phase of a selected ``spectral line" features from the (whitened) output of the Caltech 40-meter interferometer. The upper two graphs show the approximately exponential decay of the 582.396 Hz violin mode; the lower two graphs show the amplitude and phase of the third harmonic of the 60Hz line noise (note the remarkable amplitude stability).

Authors: Bruce Allen (ballen@dirac.phys.uwm.edu) and Adrian Ottewill (ottewill@relativity.ucd.ie).

Comments: None.