Not able to plot with large parameters value

Question

I want to plot the beta distribution with beta.pdf(x, a, b) but I only get the following error:

FloatingPointError: underflow encountered in exp.

This is my code:

x = np.linspace(0, 1, 990)
a=50.5
b=940.5

y = beta.pdf(x, a, b)

plt.title("PDF of Beta (Bell-shape)", fontsize=20)
plt.xlabel("X", fontsize=16)
plt.ylabel("Probability Density", fontsize=16)
plt.plot(x, y, linewidth=3, color='firebrick')
plt.annotate("Beta(50.5,940.5)", xy=(0.15, 2), size = 100, ha='center', va='center', color='firebrick')

fig = plt.gcf()
plt.show()


fig.savefig('my_plot.png',dpi=130)

Works for me. Can you include the versions of the different modules? — DavidG, May 06 '20 at 09:40
how to know the version of the modules. Can u please help me as it works for u but not for me. — Mohd Bilal, May 06 '20 at 09:46
I am assuming you are using scipy.stats? If so you can find scipy version using `print(scipy.__version__)` — DavidG, May 06 '20 at 09:48
OK, its probably your numpy then. What version is that? `print(np.__version__)` — DavidG, May 06 '20 at 10:03
I updated some packages. Now it is working perfectly for me as well. Thank you @DavidG — Mohd Bilal, May 06 '20 at 10:51
Which packages did you update exactly? I'm facing the same problem right now. — Hagbard, May 12 '20 at 14:11
I was using anaconda that was not the updated version. So, first, I update my anaconda with `conda update conda` then I update all the packages with `conda update --all`. then above implementation works for me. Hope it helps. — Mohd Bilal, May 12 '20 at 14:29

score 1 · Answer 1 · answered May 12 '20 at 15:48

I had the exact same problem while trying to create a seaborn violin plot. Thus, I updated the packages scipy, matplotlib, numpy and seaborn with the following commands:

pip install --upgrade scipy
pip install --upgrade matplotlib
pip install --upgrade numpy
pip install --upgrade seaborn

Unfortunately, that didn't help either, so I ended up using the seterr method of numpy to simply ignore this warning (as described in this old answer). You can set it as follows at the beginning of your code:

numpy.seterr(under='ignore')

Not able to plot with large parameters value

1 Answers1