Andy Long’s paper on machine learning and design of digital colloid information storage elements published in Soft Matter

Andy’s paper “Nonlinear machine learning and design of reconfigurable digital colloids” in collaboration with Eric Jankowski (Boise State) and Carolyn Phillips (Neurensic Inc.) has been published in Soft Matter (http://dx.doi.org/10.1039/C6SM01156J). So-called “digital colloids” — a cluster of freely rotating “halo” particles tethered to the surface of a central particle — present a novel ultra-high density soft memory storage substrate in which information is encoded in the arrangement of halo particles around the central colloid. Information is written by reconfiguring the halo particle arrangement, and information read by interrogating the particular rotationally distinguishable halo particle state. A digital colloid comprising 6 halo particles can store an estimated 1.1 TB/g (1.4 TB/cm3). Using diffusion map nonlinear manifold learning to discover the low dimensional free energy surface governing the halo particle collective reconfigurations and mean first passage time calculations to define interconversion rates between states, we characterized the thermodynamics, morphology, and kinetics of the structural information storage process. Using this framework we designed a digital colloid with particular central to halo particle diameter to trade off the competing design criteria of memory addressability and volatility. Well done, Andy!