Class Of Particle Filters Research Articles

On the performance of particle filters with adaptive number of particles

We investigate the performance of a class of particle filters (PFs) that can automatically tune their computational complexity by evaluating online certain predictive statistics which are invariant for a broad class of state-space models. To be specific, we propose a family of block-adaptive PFs based on the methodology of Elvira et al. (IEEE Trans Signal Process 65(7):1781–1794, 2017). In this class of algorithms, the number of Monte Carlo samples (known as particles) is adjusted periodically, and we prove that the theoretical error bounds of the PF actually adapt to the updates in the number of particles. The evaluation of the predictive statistics that lies at the core of the methodology is done by generating fictitious observations, i.e., particles in the observation space. We study, both analytically and numerically, the impact of the number K of these particles on the performance of the algorithm. In particular, we prove that if the predictive statistics with K fictitious observations converged exactly, then the particle approximation of the filtering distribution would match the first K elements in a series of moments of the true filter. This result can be understood as a converse to some convergence theorems for PFs. From this analysis, we deduce an alternative predictive statistic that can be computed (for some models) without sampling any fictitious observations at all. Finally, we conduct an extensive simulation study that illustrates the theoretical results and provides further insights into the complexity, performance and behavior of the new class of algorithms.

A stable particle filter for a class of high-dimensional state-space models

Abstract We consider the numerical approximation of the filtering problem in high dimensions, that is, when the hidden state lies in ℝd with large d. For low-dimensional problems, one of the most popular numerical procedures for consistent inference is the class of approximations termed particle filters or sequential Monte Carlo methods. However, in high dimensions, standard particle filters (e.g. the bootstrap particle filter) can have a cost that is exponential in d for the algorithm to be stable in an appropriate sense. We develop a new particle filter, called the space‒time particle filter, for a specific family of state-space models in discrete time. This new class of particle filters provides consistent Monte Carlo estimates for any fixed d, as do standard particle filters. Moreover, when there is a spatial mixing element in the dimension of the state vector, the space‒time particle filter will scale much better with d than the standard filter for a class of filtering problems. We illustrate this analytically for a model of a simple independent and identically distributed structure and a model of an L-Markovian structure (L≥ 1, L independent of d) in the d-dimensional space direction, when we show that the algorithm exhibits certain stability properties as d increases at a cost 𝒪(nNd2), where n is the time parameter and N is the number of Monte Carlo samples, which are fixed and independent of d. Our theoretical results are also supported by numerical simulations on practical models of complex structures. The results suggest that it is indeed possible to tackle some high-dimensional filtering problems using the space‒time particle filter that standard particle filters cannot handle.

Particle Filters for nonlinear data assimilation in high-dimensional systems

Particle Filters are Monte-Carlo methods used for Bayesian Inference. Bayesian Inference is based on Bayes Theorem that states how prior information about a system, encoded in a probability density function, is updated when new information in the form of observations of that system become available. This process is called data assimilation in the geosciences. This contribution discusses what particle filters are and what the main issue is when trying to use them in the geosciences, in which the data-assimilation problem is typically very high dimensional. An example is numerical weather forecasting, with a state-space size of a billion or more. Then it discusses recent progress made in trying to beat the so-called 'curse of dimensionality', such as localisation and clever ways to slightly change the model equations to obtain better approximations to the posterior probability density via so-called proposal densities. This culminates in a new class of particle filters that is indeed able to provide estimates of the posterior probability density. The emphasis is not on mathematical rigour but on conveying the main new ideas in this rapidly growing field.

State estimation and prediction using clustered particle filters

Particle filtering is an essential tool to improve uncertain model predictions by incorporating noisy observational data from complex systems including non-Gaussian features. A class of particle filters, clustered particle filters, is introduced for high-dimensional nonlinear systems, which uses relatively few particles compared with the standard particle filter. The clustered particle filter captures non-Gaussian features of the true signal, which are typical in complex nonlinear dynamical systems such as geophysical systems. The method is also robust in the difficult regime of high-quality sparse and infrequent observations. The key features of the clustered particle filtering are coarse-grained localization through the clustering of the state variables and particle adjustment to stabilize the method; each observation affects only neighbor state variables through clustering and particles are adjusted to prevent particle collapse due to high-quality observations. The clustered particle filter is tested for the 40-dimensional Lorenz 96 model with several dynamical regimes including strongly non-Gaussian statistics. The clustered particle filter shows robust skill in both achieving accurate filter results and capturing non-Gaussian statistics of the true signal. It is further extended to multiscale data assimilation, which provides the large-scale estimation by combining a cheap reduced-order forecast model and mixed observations of the large- and small-scale variables. This approach enables the use of a larger number of particles due to the computational savings in the forecast model. The multiscale clustered particle filter is tested for one-dimensional dispersive wave turbulence using a forecast model with model errors.

Implicit equal‐weights particle filter

Filter degeneracy is the main obstacle for the implementation of particle filters in nonlinear high‐dimensional models. A new scheme, the implicit equal‐weights particle filter (IEWPF), is introduced, in which samples are drawn implicitly from proposal densities with a different covariance for each particle, such that all particle weights are equal by construction.We test and explore the properties of the new scheme using a 1000 dimensional simple linear model and the 1000 dimensional nonlinear Lorenz96 model and compare the performance of the scheme with that of a local ensemble transformed Kalman filter (LETKF). The new scheme is never degenerate and shows good and consistent performance in all experiments. The LETKF has lower root‐mean‐square errors at observed grid points, but its ensemble spread is too low at unobserved grid points, where the IEWPF performs better. Furthermore, the IEWPF has a consistent spread in all experiments.This new filter opens up a new class of particle filters that, by construction, do not suffer from the curse of dimensionality.

Acoustic source tracking in long baseline microphone arrays

Long baseline microphone arrays, where the distance between single microphones is large, is a cost-effective method to monitor activity in a large region of interest. The high range between source and sensors in a long baseline setup, generally leads to low signal-to-noise ratios (SNR). This affect the performance of localization. This paper investigates how energy based localization (EBL) can be used for localization and tracking of both single and multiple acoustic sources, within a long baseline array. Least-Squares (LS) optimization have been used for EBL-based localization, however the localization performance is sensitive to low SNR. We propose a tracking scheme based on a cost reference particle filter (CRPF) to increase performance during low SNR. The CRPF is a new class of particle filters which is able to estimate the system state from the available observations without a priori knowledge of any probability density function. We present a modified cost function, for EBL based localization, which is incorporated in the CRPF framework to increase tracking performance during simulated wind gusts and background noise. The proposed method outperforms LS based localization in the case of low SNR.

Optimality of CUSUM Rule Approximations in Change-Point Detection Problems: Application to Nonlinear State–Space Systems

The well-known cumulative sum (CUSUM) sequential rule for abrupt model change detection in stochastic dynamic systems relies on the knowledge of the probability density functions of the system output variables conditional on their past values and on the system functioning mode at each time step. This paper shows how to build an asymptotically optimal detection rule under the common average run length (ARL) constraint when these densities are not available but can be consistently estimated. This is the case for nonlinear state-space systems observed through output variables: for such systems, a new class of particle filters based on convolution kernels allows to get consistent estimates of the conditional densities, leading to an optimal CUSUM-like filter detection rule (FDR).