Computer Science – Information Theory
Scientific paper
2011-06-25
Computer Science
Information Theory
Submitted
Scientific paper
In this paper we study certain properties of R\'{e}nyi entropy functions $H_\alpha(\mathcal{P})$ on the space of discrete probability distributions with infinitely many probability masses. We prove some properties that parallel those known in the finite case. Some properties on the other hand are quite different in the infinite case, for example the (dis)continuity in $\mathcal{P}$ and the problem of divergence and behaviour of $H_\alpha(\mathcal{P})$ at the point of divergence. Finally, we prove that, given a sequence of distributions $\mathcal{P}_n$ converging to $\mathcal{P}$ with respect to the total variation distance, $\lim_{\alpha\to1+}\lim_{n\to\infty} H_\alpha(\mathcal{P}_n)$ is in general not equal to $\lim_{n\to\infty}\lim_{\alpha\to1+} H_\alpha(\mathcal{P}_n)$, so interchanging limiting operations (which is often done in applications) is not justified in this case.
Kovačević Mladen
Šenk Vojin
Stanojević Ivan
No associations
LandOfFree
On Rényi Entropies Over Countably Infinite Alphabets does not yet have a rating. At this time, there are no reviews or comments for this scientific paper.
If you have personal experience with On Rényi Entropies Over Countably Infinite Alphabets, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and On Rényi Entropies Over Countably Infinite Alphabets will most certainly appreciate the feedback.
Profile ID: LFWR-SCP-O-144859