Parallel Non-Monte Carlo Transient Noise Simulation

A novel parallel non-Monte Carlo (MC) transient noise analysis method for general nonlinear analysis is presented. Non-MC-based transient noise methods avoid expensive and lengthy MC iterations by describing the nonlinear system with a linear time-varying (LTV) system that can be used to calculate the variance of the waveform at every time point. However, the main bottleneck in these methods is that the LTV system is dense, strongly coupled, and significantly larger compared to the original system. To address this issue, the proposed method decouples the LTV system into homogeneous and inhomogeneous solutions that can be solved independently. The inhomogeneous solution is solved in parallel by partitioning the time interval into independent subintervals, while the complexity of the homogeneous system can be decreased significantly due to the linearity of the equations.