A potential implicit particle method for high-dimensional systems
A potential implicit particle method for high-dimensional systems
复制标题
高维系统的潜在隐式粒子方法
DOI：
10.5194/npg-20-1047-2013
复制
发表时间：
2013
影响因子：
2.2
通讯作者：
Y. Spitz
中科院分区：
文献类型：
--
作者：
B. Weir;Robert N. Miller;Y. Spitz
本文提出了一种用于高维状态估计的粒子方法。该方法不是通过随机预测与给定观测值的距离来加权随机预测，而是基于观测值对最佳解周围的粒子系综进行采样（即，这是隐含的）。它与其他隐式方法不同，因为它包括前一同化时间的状态作为最优解的一部分（即，它是滞后平滑器）。这是通过使用一个混合模型的背景discovery的前一个状态。在高维线性高斯示例中，基于混合的隐式粒子平滑器不会塌陷。此外，仅使用少量的粒子，隐式方法能够检测在两个非线性，多维广义的双阱中的跃迁。添加一个将采样分布训练到目标分布的步骤可以防止过渡期间的崩溃，这是强非线性事件。为了产生类似的估计，其他方法需要更多的粒子。
This paper presents a particle method designed for high-dimensional state estimation. Instead of weighing ran- dom forecasts by their distance to given observations, the method samples an ensemble of particles around an opti- mal solution based on the observations (i.e., it is implicit). It differs from other implicit methods because it includes the state at the previous assimilation time as part of the optimal solution (i.e., it is a lag-1 smoother). This is accomplished through the use of a mixture model for the background dis- tribution of the previous state. In a high-dimensional, lin- ear, Gaussian example, the mixture-based implicit particle smoother does not collapse. Furthermore, using only a small number of particles, the implicit approach is able to detect transitions in two nonlinear, multi-dimensional generaliza- tions of a double-well. Adding a step that trains the sampled distribution to the target distribution prevents collapse during the transitions, which are strongly nonlinear events. To pro- duce similar estimates, other approaches require many more particles.