Compact Probabilistic Modeling of Voltage Controlled MTJ Towards Efficient Ising Machines
Year: 2026
Type: conference-paper
Abstract: Emerging probabilistic computing paradigms based on Ising machines require energy-efficient and scalable hardware implementations. While Voltage-Controlled Magnetic Anisotropy (VCMA) Magnetic Tunnel Junctions (MTJs) are promising candidates for probabilistic bits (p-bits), their stochastic dynamics, governed by the stochastic Landau-Lifshitz-Gilbert (s-LLG) equation, pose significant challenges for efficient circuit-level simulation. To address this bottleneck, we propose a physics-based compact probabilistic model for VCMA-MTJs. Instead of computationally expensive numerical integration, this model utilizes an analytical damped sinusoidal function to precisely capture the voltage-dependent switching probability and stochastic incubation delay. Calibrated against experimental foundry data, the proposed model achieves a simulation speedup of approximately$75 \times$compared to traditional numerical approaches. System-level validations in standard circuit simulators demonstrate the generation of high-quality uncorrelated stochastic bit-streams and the successful resolution of a visual pattern recovery problem. This work bridges the gap between device physics and system design, providing a robust tool for the co-simulation of large-scale VCMA-based probabilistic computing systems.
Language: English
Subfield: Condensed Matter Physics
Field: Physics and Astronomy
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7207581998