Оптимізація використання пам’яті при експоненційному зростанні багатовимірної моделі пасажиропотоку з нечіткими параметрами

Abstract

The paper investigates the problem of exponential growth in memory consumption arising in passenger flow modeling on urban routes using fuzzy sets. It is shown that forming the state set at each computation step as a Cartesian product leads to a multiplicative increase in model dimensionality and rapid exceeding of acceptable RAM capacity. The main approaches to reducing memory consumption are analyzed, including aggregation of identical values, threshold pruning of low-significance states, discretization and quantization, the use of sparse data structures, reduction of intermediate computation layers, and tensor decompositions. It is demonstrated that applying these methods significantly reduces the required memory volume and enables practical implementation of fuzzy passenger flow models under limited computational resources.