This fascinating phenomenon from Sid Nagel’s group in U Chicago, on memory and noise seems like a continuous analogue to Josh Bacigalupi's model.
Some non-equilibrium systems can store information of their external driving in an unexpected manner. They “learn” multiple driving amplitudes that can subsequently be read out. Notably, only one memory is retained after many driving cycles, even if all of the amplitudes are continually fed in. If noise is added, the system can store all memories indefinitely. While exceedingly counterintuitive, these properties can be understood from simple considerations, in the case of a simple model of sheared viscous suspensions (originally developed by Corte et al. to model the reversibility-irreversibility transition in these systems---see L. Corte et al., Nature Phys. 4, 420, 2008). Furthermore, the memory phenomenon is expected to be generic—the same effect is seen in simulations and experiments on traveling charge-density waves. We are exploring a variety of systems with both simulation and experiment, to understand the sufficient conditions for these memories to exist.
