Chen, B., Hou, Y., Gan, C. K., & Zeng, M. (2023). Micromagnetic realization of energy-based models using stochastic magnetic tunnel junctions. Applied Physics A, 129(9). https://doi.org/10.1007/s00339-023-06931-4
Abstract:
Energy-based models (EBMs) can bridge physics, machine learning, and statistics. EBMs provide a unified and powerful platform to describe, learn, and optimize complex systems. In this paper, we propose a neuromorphic implementation of EBMs using a network of stochastic magnetic tunnel junctions (MTJs) that can perform energy minimization and solve optimization problems. Our implementation builds on the Object Oriented MicroMagnetic Framework (OOMMF). We derive the different energy terms and map them to the micromagnetic Landau-Lifshitz-Gilbert (LLG) equation. We then develop a C + + module for EBMs which integrates seamlessly with OOMMF. We demonstrate our implementation on a full set of logic gates using stochastic MTJs networks. Our method offers several advantages, including fast modeling of EBMs with spintronic devices and design insights for stochastic MTJ-based neuromorphic circuits.
License type:
Publisher Copyright
Funding Info:
This research / project is supported by the A*STAR - Career Development Fund
Grant Reference no. : C210812054
Description:
This version of the article has been accepted for publication, after peer review and is subject to Springer Nature’s AM terms of use, but is not the Version of Record and does not reflect post-acceptance improvements, or any corrections. The Version of Record is available online at: http://dx.doi.org/10.1007/s00339-023-06931-4