how to fix axes position by using set_position method in python matplotlib?

Question

I think this is quite easy but I searched the internet and matplotlib users mailing list and not able to find an answer. ax2 is an inset axes within the "ax" axes in figure "fig", which I make by following here: http://matplotlib.sourceforge.net/examples/pylab_examples/axes_demo.html

but now my problem is that I cannot fix the ax2 the exact position I want, it seems that draw() command change this:

In [352]:
ax2.set_position([0.125,0.63,0.25,0.25])

In [353]:
ax2.get_position()

Out[353]:
Bbox(array([[ 0.125,  0.63 ],
       [ 0.375,  0.88 ]]))

In [354]:
draw()

In [355]:
ax2.get_position()

Out[355]:
Bbox(array([[ 0.15625,  0.63   ],
       [ 0.34375,  0.88   ]]))

notice that, after "draw()" command, the x0 of ax2 changed. could anyone give any hints?

thanks!

Solution

Try setting the position by specifying the parameters as a Bbox object:

>>> ax2.set_position(matplotlib.transforms.Bbox(array([[0.125,0.63],[0.25,0.25]])))
>>> ax2.get_position()
Bbox(array([[ 0.125,  0.63 ],
       [ 0.25 ,  0.25 ]]))
>>> draw()
>>> ax2.get_position()
Bbox(array([[ 0.125,  0.63 ],
       [ 0.25 ,  0.25 ]]))

With this you can see the settings are round-tripping like you would expect.

In addition, you could instead look at the results as a tuple like so:

>>> a.set_position([0.125,0.63,0.25,0.25])
>>> a.get_position().bounds
(0.125, 0.63, 0.25, 0.25)

It just depends if you want to look at the location/position as a bounding box: x0,y0 by x1,y1, or instead by a location and size: x,y,width,height.

Currently you are telling it to set a x,y,width,height, then it's telling you the x0,y0,x1,y1.