A Non-Linear Partial Differential Equation Solver Based on Analog Memristor Crossbar Arrays
An analog memristor-crossbar-arraybased non-linear partial differential equation (PDE) solver is developed and verified through circuit-level simulations using realistic memristor device data. Several methodologies are suggested for the matrixvector multiplication (MVM) and weight update (WU) operations, which are the essenses of the highly parallelised PDE solver. An example of solving the Burgers’ equation is demonstrated. Results suggest that, in comparison with traditional computers, the proposed memristor PDE solver offers dramatic performance improvement and energy savings.
Authors
- Nuo Xu (ORCID: https://orcid.org/0000-0003-1224-7345)
- Chao Du (ORCID: https://orcid.org/0000-0002-0438-1416)
- Jiantao Zhou (ORCID: https://orcid.org/0000-0002-6015-2618)
- Jiabao Ye (ORCID: https://orcid.org/0009-0009-3406-6714)
- Wei Lü (ORCID: https://orcid.org/0000-0003-4731-1976)
- Bing Chen (ORCID: https://orcid.org/0000-0001-5284-8618)
- Rongjun Mu
Institutions
- University of Michigan (US)
- Siemens (United States) (US)
- Zhejiang University (CN)
Publication Details
- Journal
- IEEE Journal of the Electron Devices Society
- Published
- 2025-01-01
- DOI
- https://doi.org/10.1109/jeds.2025.3573409
- Citations
- 1
- Primary Topic
- Neural Networks Stability and Synchronization
- Type
- article
- Field-Weighted Citation Impact
- 1.49
Funders
- National Natural Science Foundation of China
- Natural Science Foundation of Zhejiang Province