Title:Optimal Construction of Basis Functions for Projection States in Discrete Multicomponent System

Abstract:Our recent study reveals that macroscopic structure in thermodynamically equilibrium state and its temperature dependence for classical descrete system can be well-characterized by a single specially-selected microscopic state (which we call projection state: PS), whose structure can be known a priori without any information about energy or temperature. Although PS can be universally constructed for any number of components R, practical application of PS to systems with R >= 3 is non-trivial compared with R = 2 (i.e., binary system). This is because multicomponent system should inevitablly requires linear transformation from conventinal basis functions to intuitively-interpreted cluster probability basis, whose explicit formula for the transformation has been provided only for pair clusters in ternary (R = 3) system. We here propose strategy to directly determine probability for likeand unlike-atom pair consisting of a chosen single element by a single PS, whose minimal condition of microscopic structure is provided. This can be done by providing complete relationships between basis functions and pair cluster probability, which avoids linear transformation of basis functions, leading to decrease of accumulating predictive error for macroscopic structure in equilibrium state based on information about PS.