Monte Carlo Approximations for General State-Space Models
针对非线性非高斯状态空间模型，提出了两种通过简单拒绝算法顺序生成滤波密度和平滑密度样本的方法，并在多个例子中与其他方法比较。
Abstract Nonlinear and non-Gaussian state-space models form a large and flexible model class in time series analysis. Two methods for sequentially generating samples from filter densities and smoother densities by simple rejection algorithms are introduced. We illustrate the behavior of our methods in several nonlinear and non-Gaussian examples and compare them with other well-known methods. Key Words: Kalman filterKalman smootherMonte Carlo methodsNonlinear time series analysisRobustness for time seriesState-space models