Research Paper#Circuit Simulation, Physics-Informed Neural Networks, NeuroSPICE🔬 ResearchAnalyzed: Jan 3, 2026 18:35
NeuroSPICE: Physics-Informed Neural Networks for Circuit Simulation
Analysis
This paper introduces NeuroSPICE, a novel approach to circuit simulation using Physics-Informed Neural Networks (PINNs). The significance lies in its potential to overcome limitations of traditional SPICE simulators, particularly in modeling emerging devices and enabling design optimization and inverse problem solving. While not faster or more accurate during training, the flexibility of PINNs offers unique advantages for complex and highly nonlinear systems.
Key Takeaways
- •NeuroSPICE utilizes Physics-Informed Neural Networks (PINNs) for circuit simulation.
- •It solves circuit differential-algebraic equations (DAEs) by minimizing the residual through backpropagation.
- •Offers advantages in modeling emerging devices, design optimization, and inverse problems.
- •Provides a flexible approach for simulating highly nonlinear systems.
Reference
“NeuroSPICE's flexibility enables the simulation of emerging devices, including highly nonlinear systems such as ferroelectric memories.”