This is an outdated version published on 2024-09-30. Read the most recent version.

COMPUTATIONAL MODELING OF MEMRISTOR NETWORKS: INFLUENCE OF CONECTIVITY AND ENVIRONMENTAL CONDITIONS ON THEIR ELECTRICAL RESPONSE

Authors

J. Furlanetto Instituto Sabato, Universidad Nacional de San Mart´ın – Comisi´on Nacional de Energ´ıa At´omica, Avda. General Paz 1499, C1650 San Mart´ın, Prov. de Buenos Aires
R. Weht 1 Instituto Sabato, Universidad Nacional de San Mart´ın – Comisi´on Nacional de Energ´ıa At´omica, Avda. General Paz 1499, C1650 San Mart´ın, Prov. de Buenos Aires 2Departamento F´ısica de la Materia Condensada, Comisi´on Nacional de Energ´ıa At´omica – CONICET, Avda. General Paz 1499, C1650 San Mart´ın, Prov. de Buenos Aires
C. Quinteros Instituto de Ciencias F´ısicas (ICIFI), Universidad Nacional de San Mart´ın - CONICET, 25 de Mayo y Francia, C1650 San Mart´ın, Provincia de Buenos Aires

Abstract

With the perspective of implementing neuromorphic devices using memristor networks, we present a numericalcomputational platform that allows studying their electrical transport properties. It consists of a function that describes the behavior of an individual memristor (which enables the consideration of an experimental case of interest) and a
connection diagram between memristors. Using this platform, two-dimensional networks of various sizes are studied,
also incorporating environmental effects (such as humidity and temperature). Furthermore, by selecting the electrically
accessible ports, the impact of connectivity on the electrical properties of the network is evaluated. The results indicate
that, beyond the individual characteristics of the memristors (determined by the particular selected model), the degree of
connectivity, the topology of the connections, and the environmental conditions significantly influence the macroscopic
electrical response of the networks.