Research Projects

Our proposed project aims to create a theoretical and computational framework for neuromorphic computing (NC) using spin-torque nano-oscillators (STNOs) and Spin-all nano-oscillators. The Landau equations govern these nanoscale magnetic devices, the Landau-Lifshitz-Gilbert-Slonczewski (LLGS) equations, and exhibit rich nonlinear dynamics, including multistability, synchronization, and tunable oscillatory behavior. The core hypothesis of this work is that such STNOs, when interconnected via bilinear interlayer exchange coupling and driven by spin-transfer torque (STT), can emulate spiking neurons and synaptic connectivity, enabling brain-like computation of Neuromorphic Computation (NC). Building on our previous work that demonstrated multistability and high-frequency dynamics in tilted polarized and bilinear-coupled STNOs (Phys. Rev.. B 107, 224434 (2023); J. Appl. Phys. 131, 243902 (2022)), the proposed study aims to transform these devices into functional elements of a neuromorphic platform. The STNOs will be treated as excitable spiking neurons, and their coupling strengths will be dynamically modulated using Hebbian and STDP-inspired rules, allowing for real-time learning and pattern recognition. Our methodology involves focusing on the LLGS equations and modeling many STNO arrays with adaptive coupling, identifying spiking and synchronization regimes through bifurcation and phase space analysis, and simulating network behavior in response to external input patterns. Benchmark neuromorphic tasks such as binary classification, winner-take-all decision making, and associative memory recall will be implemented. Furthermore, the project aims to demonstrate STNO-based reservoir computing (RC) using the system's high-dimensional transient response to process temporal signals. We also plan to illustrate coupling schemes in STNO arrays to achieve synchronization for low-power neuromorphic architectures. A key innovation is the Weighted Spin-Torque Nano-Oscillator (WSTNO) circuit, which connects STNOs via a resistive network to optimize power output and enable tunable frequency enhancement. We will also examine angular-dependent spin torque amplitudes—both in-plane and out-of-plane in tilted polarizer STNOs to expand frequency tunability under zero external fields. All studies will include thermal effects, field-related torque, and time delay. If successful, this project will bridge the gap between nonlinear spintronic dynamics and neuromorphic computing, offering an energy-efficient, scalable hardware substrate for neuromorphic systems. The theoretical advances will provide new models of computation rooted in physical oscillator dynamics, and the simulations will lay the groundwork for future experimental realization of STNO-based intelligent systems.