Rate of Entropy Production as a Physical Selection Principle for Mode-Mode Transitions in Non-Equilibrium Systems: With an Application to a Non-Algorithmic Dynamic Message Buffer

We examine a generic set of amplitude equations proposed earlier by Haken that describes the emergence and bifurcations of modes and spatio-temporal patterns of selforganizing non-equilibrium systems. We relate feedback parameters occurring in the amplitude equations to pumping processes associated with entropy production. In doing so, we show that the rate of entropy production determines which mode-mode transitions are allowed and which not. Roughly speaking, transitions occur from modes of low rate of entropy production towards modes of high rate of entropy production (selection principle). In line with the recent efforts in the field of physical intelligence, we outline how physical, non-algorithmic, self-organizing systems satisfying Haken’s amplitude equations may be used to design a dynamic input-output message buffer. The functioning of such a dynamic buffer again follows the aforementioned selection principle: the buffer switches between input and output modes in order to select modes with relatively high rates of entropy production. Moreover, only mode-mode transitions are allowed that increase the rate of entropy production of the active mode.