plot_evoked_joint accepts a times parameter:

I think the following changes would be an improvement:

• make it possible to determine the number of evenly spaced topos via "auto" (currently hardcoded to 5) and "peaks" (currently hardcoded to 3)
• make it possible to determine a window (or one window per peak) in which "peaks" should be found (currently hardcoded to full time window)

suggested API:

• accept a dict mapping a single str key to value(s):
• if "auto", must be either:
  • an integer bigger than 1, corresponding to the number of evenly spaced topos to be plotted
  • a tuple of the form (n, (tmin, tmax)) to plot n evenly spaced topos in the window (tmin, tmax)
• if "peaks", must be either:
  • an integer bigger than 1, corresponding to the number of peaks (over full time window) to be plotted
  • a tuple of tuples, where each tuple refers to a (tmin, tmax) of a window in which to plot the peak of that window

For example

times = "auto"  # will plot 5 evenly spaced peaks over the whole window
times = np.linspace(0.2, 0.6, 3)  # will plot 3 evenly spaced peaks between 0.2 and 0.6

times = {"peaks": 1}  # will plot 1 single peak within the full time window
times = {"peaks": ((0, 0.1))}  # will plot single peak within 0 to 0.1s window
times = {"peaks": ((0, 0.1), (0.5, 0.8))}  # will plot two peaks within their windows

times = {"mykey": 5}  # error, only "peaks" is an accepted keys

From just thinking about it (haven't looked at the code yet), this should be fairly straight forward, and helpful.

WDYT?