Simulation of memristive computing architectures with realistic device characteristics

Nyomtatóbarát változatNyomtatóbarát változat
Doctoral school: 
Fizikai Tudományok Doktori Iskola
Halbritter András Ernő
Department of Physics
Job title: 
Academic degree: 
Doctor of the Hungarian Academy if Sciences (DSc)

In the recent decade resistive switching phenomenon was established in various material systems. These, so-called memristive devices promote novel, energy efficient, fast and compact technologies in the fields of nonvolatile data storage, in-memory computing, the hardware implementation of artificial neural networks, probabilistic optimization or autonomous signal detection. All these applications, however, strongly rely on the material properties of the involved memristive system, like the device noise, or the dynamical characteristics.

Our research group performs a broad range of characterization measurements on various memristive systems, including 1/f-noise analysis, or the investigation of the switching dynamics and the characteristic physical time-scales of the devices down to the nanosecond range. The PhD candidate will perform computer simulations on larger-scale memrsitive architectures, investigating the sensitivity of the computing efficiency on the realistic, experimentally determined operation characteristics of the individual devices. A primary goal is the investigation of memristive Hopfield neural networks, which are capable of solving complex optimization problems in a highly energy-efficient manner. In these architectures, the device noise is not necessarily a disturbing factor, but a properly tailored, and well-tunable noise level may be harvested as a computational resource. Later, the investigations will be extended towards further computing schemes, like oscillator networks, or various reservoir computing schemes.


Computer programming skills, some experience in data science. Experience in resistive switching experiments is an advantage.
Project type: 
PhD project for standard admission