The R package tseriesEntropy implements an entropy measure of dependence based on the Bhattacharya-Hellinger-Matusita distance.
It can be used as a (nonlinear) autocorrelation/crosscorrelation function for continuous and categorical time series.
The package includes tests for serial dependence and nonlinearity based on it.
Some routines have a parallel version that can be used in a multicore/cluster environment. The package makes use of S4 classes.
You can install the stable version on CRAN:
install.packages('tseriesEntropy')
This package is free and open source software, licensed under GPL.