How matplotlib automatically set x-axis and y-axis limits for bar graph ?

[scidam]

I didn't explore source code of matplotlib in detail, but in general, computation of values where ytick labels will be shown is performed by a matplotlib.ticker instance. By default, it is ticker.AutoLocator:

from matplotlib.ticker import AutoLocator
aul = AutoLocator()
aul.tick_values(1, 3)

Output:
array([1.  , 1.25, 1.5 , 1.75, 2.  , 2.25, 2.5 , 2.75, 3.  ])

from pylab import *
plt.plot([1,2,3])
plt.show() # yields a graph with ticks placed in above positions

This default behavior is made for convenience, and it could be easily overridden using axes.set_yticks etc. Note, there are minor and major ticks (we are talking about major one here), so underlying machinery is quite tricky.