Intelligent functions emerging from spatial correlation among phase-change-based memory and processing units

Abstract

New computing functionality emerging from spatial correlations due to the near-field interaction between local processing and memory elements is discussed. In particular, we investigate the possibility of solving a problem analogous to the spin-glass problem by using a coupled oscillator (dipole) system, in which the individual coupling strengths can be modified to optimize the system so that the exact solution can be easily obtained. For this algorithm, we propose an implementation based on a coupled plasmon-particle system interacting with a phase-change material, which exhibits a threshold behavior and plasticity to provide processing and memory functions, respectively.