[Python] resetting data limits on zoom in Matplotlib python program (long)

I've created a program to display very high sample rate data in profile. There are 6 channels (mag[0..2] & diff[0..2]) sampled at 30 kHz. I display them in 3 profiles, mag[0] with diff[0], etc. All plots share the same x axis (nptime), which is numpy timedate64, with different scales for the mag and diff. Axes 0 to 2 represent the mag profiles and 3 to 5 are diff.

In order to speed up the display I slice the data to the window limits and a desampling rate of 1, 10 or 100, based on the x extents. I do this in a callback on the xlim_changed event for axis 0.

It almost works. I call set_ylim within the callback using the data range for the windowed data. For the mag channels the y limits are set correctly for 2 of the 3 panels, but the panel I zoomed on (whichever that is) ends up displaying the zoom extents not the new extents. For the diff channels I've set up axes 3 to 5 to sharey, so I only need to set axis 3, but the set_ylim is ignored. It ALWAYS displays with the y extent from the zoom. I actually don't want the ylim for diff to change at all, I always set it to the same value.

def reset_data(self, ax):
# the viewLim extents are decimal days, so some coordinate conversion is required
width = ax.viewLim.width*86400
start=np.datetime64(pd.to_datetime(self.nptime[0]).date()) +np.timedelta64(int(ax.viewLim.min[0]%1*8.64e10),'us')
idx0 = self.find_nearest(self.nptime, start)
end=np.datetime64(pd.to_datetime(self.nptime[0]).date()) +np.timedelta64(int(ax.viewLim.max[0]%1*8.64e10),'us')
idx1 = self.find_nearest(self.nptime[idx0:], end) + idx0

# the get_rate function returns 1 of 3 desample factors 1, 10 or 100
rate = self.get_rate(width)

xdata = self.nptime[idx0:idx1:rate]
for i in range(3):
ydata = self.mag[i, idx0:idx1:rate]
ymin = ydata.min()
ymax = ydata.max()
diff = (ymax - ymin)*.1

# self.lines is an array of the plots returned by self.ax[].plot()
self.lines[0].set_data(xdata,ydata)
self.ax.set_ylim(ymin-diff, ymax+diff)
for i in range(3):
ydata = self.diff[i, idx0:idx1:rate]
self.lines[i+3][0].set_data(xdata,ydata)
self.ax[3].set_ylim(self.min_diff, self.max_diff)


before zoom:

after zoom:

Note that the panel limits for the mag (blue) channels have been set to the data limit (+10%) for panels 2&3, but the extents of the zoom box on panel 1. Also the pink trace is 7.5 to 27 on all 3 panels, which represents the zoom extents, not the limits I set.

Sorry for the long post. I can't explain it any more briefly.

Continue reading...