Anderson localization in transition metal oxides and its application in neuromorphic circuitry

It has become increasingly clear that the performance of modern computing architectures is reaching a quantum mechanical road-block, and that neuromorphic computing, a brain inspired computing model, is a very promising paradigm in going beyond the von Neumann computing architecture. At the very core of this technology, is the ability to induce fast, reversible metal-insulator transitions. We propose that Anderson localization in select transition metal oxides can be exploited to design a new generation of memristors. Utilizing toy models as well as first principles descriptions of niobium and vanadium oxide structures, we show that disordering the system can be leveraged to engineer the metal insulator switching, a key physical mechanism in the construction of modern day neuromorphic circuits. Furthermore, we demonstrate that a simple metric of eigenstate localization, the Gini vector, can be an efficient way to capture the localization properties in a first-principles tight binding model.