k)
time series and we wish to investigate
the effect of relocating the origin
of each series on their cross-correlations.
This is similar to a crude algorithm for
co-registration of images in which the correlation
between two images is computed for all "possible"
offsets (x,y) for the "second" image.
In the same way, if the time origin for a time series
is an unknown parameter, it can be estimated by
optimizing the correlation as a function of the origin.
Rather than simply computing the correlations for all possible origins for all possible time series and selecting the optimal value, we provide an interactive GUI which allows the user to move each time series using a scrollbar and to observe the change in the correlation coefficients.
While the user is viewing the output, a second thread can be computing the values of the correlations for the currently active time series within a given local window so that the computations can appear greatly speedened for the user.
Have to rethink what we mean for sequential tasks in an application rather than what the user thinks as sequential. Menu Bar
Plotting Area with the time series displayed. The scale is complicated here. Do we show all possible values or just the common area. This can also be specified by the user using the Magstep option in the Data Menu.
A scrollbar for each time series which controls the position of the relevant series.