# Entropy of 2-spikes Distribution

The Entropy of the following distribution is $- \infty$

$p(x) = \frac{\delta(x=-1) + \delta(x=1)}{2}$

Mathematically the reason is because of having a $- \infty \over 2$ density probability for each of two spikes. What is the conceptual reason for having that Entropy value?

The Entropy measure: $H(x) = - \int_{-\infty}^{+\infty} p(x).ln(p(x)).dx$

-
 You need to describe what measure of entropy you are using: Shannon entropy does not produce $- \infty$ here – Henry Jun 19 '12 at 19:47