Extension of Gatlin's Informational Divergence to Markovian Stochastic Processes.

In the information theoretic framework, to some extent, the complexity ofa system can be measured by its informational divergence which is thedifference between the maximum possible value of its entropy and theactual value of the latter. In her analysis of the DNA chain, Gatlinput in evidence two different divergences. One which drops the mutualdependences between the bases (A, C, G, T) and the other one whichexplicitly refers to this dependence via conditional entropies, and thus,provides a measure of the structural complexity of the system. One showshow the explicit form of this structural divergence can be obtained fordiscrete and continuous Markovian stochastic processes, and in the lattercase, as expected, this divergence is invariant under a transformation ofvariables.