Fill area under curve in matlibplot python on log scale

Question

I'm trying to fill the area under a curve with matplotlib. The script below works fine.

import matplotlib.pyplot as plt
from math import sqrt
x = range(100)
y = [sqrt(i) for i in x]
plt.plot(x,y,color='k',lw=2)
plt.fill_between(x,y,0,color='0.8')
plt.show()

However if I set the y-scale to logarithmic (see below). It sometimes fills the area above the curve ! Can anyone help me? I would like to fill the area between the curve and y = 0.

x = range(100)
y = [sqrt(i) for i in x]
plt.plot(x,y,color='k',lw=2)
plt.fill_between(x,y,0,color='0.8')
plt.yscale('log')
plt.show()

Thanks in advance!

Answer 1

With a logarithmic y-scale, fill_between(x, y, 0) tells matplotlib to fill the region between log(0) = -infinity and log(y) . Naturally, it balks. You can avoid the problem by changing 0 to some small number like 1e-6 .

Answer 2

As mentioned, 0 -> -inf in a log scale. Thus, any plotted value that was less than or equal to zero would be problematic (requiring an infinite ylim in log space). This problem exists independently of whether you are using fill_between() or not.

Fortunately, matplotlib provides a way to handle this nicely. In the default behavior, matplotlib masks the values of every value less than or equal to zero. In your example, this means that your entire y=0 line is masked and excluded from the polygon defining the filled-between area. The result is that the polygon is simply closed by drawing a line from (100,10) down and leftward to (0,0) . Another option is to clip the values. In this case, they are set to 1e-300 and are not consulted when determining the ylim of the plot. So to get your desired result, do the following:

plt.yscale('log', nonposy='clip')