Yau-YauAL: A computer tool for solving nonlinear filtering problems Volume: 1 Issue: 1
Y.Yu Wanglabel=e1][email protected], S.Shuyuan Xulabel=e2][email protected], X.Xueda Weilabel=e3][email protected], X.Xinrui Luolabel=e4][email protected], S.Stephen Shing-Toung Yaulabel=e5][email protected], S.Shing-Tung Yaulabel=e6][email protected], R.Rongling Wu t2 label=e7][email protected] · 2025
Read Yau-YauAL: A computer tool for solving nonlinear filtering problems Volume: 1 Issue: 1 by Y.Yu Wanglabel=e1][email protected], S.Shuyuan Xulabel=e2][email protected], X.Xueda Weilabel=e3][email protected], X.Xinrui Luolabel=e4][email protected], S.Stephen Shing-Toung Yaulabel=e5][email protected], S.Shing-Tung Yaulabel=e6][email protected], R.Rongling Wu t2 label=e7][email protected] (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-08976.txt.
Introduction
Yu Wang and six co-authors package a strong nonlinear filter as an easy computer tool, in a paper from 2025. The Yau-Yau filter can track hidden states in noisy complex systems, but its math has kept many would be users away.
Their tool, written in a common stats language with fast code at its core, solves the key equation by grid steps in a stable way. A point and click screen lets users tune values and watch results live, with no need to code or solve equations by hand. The open, modular build invites tweaks for fields from signals to finance and health.
Catalog record
| CREATOR | Y.Yu Wanglabel=e1][email protected], S.Shuyuan Xulabel=e2][email protected], X.Xueda Weilabel=e3][email protected], X.Xinrui Luolabel=e4][email protected], S.Stephen Shing-Toung Yaulabel=e5][email protected], S.Shing-Tung Yaulabel=e6][email protected], R.Rongling Wu t2 label=e7][email protected] |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 21 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2506-08976
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Yau-YauAL: A computer tool for solving nonlinear filtering problems Volume: 1 Issue: 1 Y.Yu Wanglabel=e1][email protected], S.Shuyuan Xulabel=e2][email protected], X.Xueda Weilabel=e3][email protected], X.Xinrui Luolabel=e4][email protected], S.Stephen Shing-Toung Yaulabel=e5][email protected], S.Shing-Tung Yaulabel=e6][email protected], R.Rongling Wu t2 label=e7][email protected] Abstract The Yau-Yau nonlinear filter has increasingly emerged as a powerful tool to study stochastic complex systems. To leverage it to a wider spectrum of application scenarios, we pack the Yau-Yau filtering ALgorithms (YauYauAL) into a package of computer software. Yau-YauAL was written in R, designed to simplify the implementation of the Yau-Yau filter for solving nonlinear filtering problems. Combining R’s accessibility with C++ (via Rcpp) for computational efficiency, YauYauAL provides an intuitive Shiny-based interface that enables real-time parameter adjustment and result visualization. At its core, YauYauAL employs finite difference methods to numerically solve the Kolmogorov forward equation, ensuring a stable and accurate solution even for complex systems. YauYauAL’s modular design and open-source framework further encourage customization and community-driven development. YauYauAL aims to bridge the gap between theoretical nonlinear filtering methods and practical applications, without requiring expertise in differential equation solving or programming, fostering its broader impact on various scientific fields, such as signal processing, finance, medicine, and biology among a long list. keywords [class=AMS] keywords † † firstpage: 1 † † lastpage: 8 † † email: e1 † † email: e2 † † email: e3 † † email: e4 † † email: e5 † † email: e6 † † email: e7 \endlocaldefs \arxiv 2506.08976 , , , , , , and t2Corresponding author. 1 Introduction In the domain of modern control theory, filtering is an essential subject that has permeated numerous fields, such as signal processing [ 5 , 29 ] , weather prediction [ 9 , 6 ] , and aerospace engineering [ 13 , 31 ] . The principal goal of filtering is to attain the ccurate estimation or forecast of a stochastic dynamical system’s state, using a set of noisy observations [ 16 , 1 ] . For real-world applications, it is crucial that these estimations or forecasts are computed iteratively and in real-time. Nonlinear filtering, in particular, has a broad spectrum of applications in military, engineering, and commercial industries [ 26 , 28 , 27 ] . The nonlinear filtering problem considered here is to determine estimated states for a given observation history of the following signal-observation model [ 1 , 16 ] : \cases{}{d}\mathbf{x}(t)=\mathbf{f}(\mathbf{x}(t))\mathrm{d}t+\mathrm{d}\mathbf{v}(t)&\mathbf{x}(0)=x_{0},\\ \mathrm{d}\mathbf{y}(t)=\mathbf{h}(\mathbf{x}(t))\mathrm{d}t+\mathrm{d}\mathbf{w}(t)\mathbf{y}(0)=0, (1) where \mathbf{x}(t)=(x_{1}(t),\ldots,x_{D}(t))^{\top}\in\mathbb{R}^{D} and \mathbf{y}(t)=(y_{1}(t),\ldots,y_{M}(t))^{\top}\in\mathbb{R}^{M} are the state and the measurement/observation vectors at time t , respectively, \mathbf{f}(\mathbf{x})=(f_{1}(\mathbf{x}),\ldots,f_{D}(\mathbf{x}))^{\top} and \mathbf{h}(\mathbf{x})=(h_{1}(\mathbf{x}),\ldots,h_{M}(\mathbf{x}))^{\top} are given vector-valued functions, \mathbf{v}\in\mathbb{R}^{D} and \mathbf{w}\in\mathbb{R}^{M} are mutually independent standard Brownian processes. From the main results of Yau and Yau [ 35 , 36 , 42 ] , the state vector \mathbf{x}(t) can be estimated from the observation vectors \{\mathbf{y}(s)\,|\,s\in[0,t]\} by solving the Kolmogorov equations. Following the introduction of the famous Kalman filter by Kalman and Bucy in the 1960s [ 18 , 19 ] , which has been widely applied across various industries, many researchers have focused on studying nonlinear filtering (NLF) theory and developing practical NLF algorithms. A key challenge in NLF is how to find the best estimate of the state from noisy observations. The optimal estimation of the state can be expressed as a minimum mean square error estimate, which is its conditional expectation based on the observation history [ 16 ] . To achieve the best estimate, one method is to directly approximate the conditional expectation, as seen in the popular extended Kalman filter (EKF) [ 16 ] , unscented Kalman filter (UKF) [ 17 ] , and ensemble Kalman filter [ 12 ] . Both EKF and UKF assume that the posterior distribution of the states is essentially Gaussian or nearly Gaussian, which can limit their use. Another approach to the NLF problem is to calculate the conditional density function of the state. For instance, the particle filter (PF) [ 10 ] uses the empirical distribution of particles to approximate the posterior density. PF is well-known for its versatility in various NLF problems, although it is not suitable for real-time implementation in high-dimensional systems. In the late 1960s, Duncan, Mortensen, and Zakai independently developed the renowned Duncan-Mortensen-Zakai (DMZ) equation for nonlinear filtering [ 8 , 22 , 43 ] , which is satisfied by the unnormalized conditional density function of the states. The DMZ equation, a stochastic differential equation, generally does not have a closed-form solution. For practical use, it is necessary to solve the DMZ equation in real-time and without extensive memory. For finite-dimensional filtering systems, solutions to the DMZ equation can be explicitly built using Lie algebra methods. However, this is only possible for a few types of systems that have finite-dimensional filters [ 2 , 40 ] . Since the DMZ equation usually lacks a closed-form solution, many mathematicians have sought good approximations. One such method is the splitting up technique, first described by Bensoussan et al. [ 3 , 4 ] and later studied in [ 11 , 25 ] . In the 1990s, Lototsky, Mikulevicius, and Rozovskii provided a recursive, time-based Wiener chaos representation for the optimal nonlinear filter [ 20 ] . However, these methods generally require that the drift and observation terms, specifically f(x) and h(x) in system (1), are bounded. Another method to solve the DMZ equation is the direct method, which is typically applicable only to Yau filtering systems where the drift term is a linear function plus a gradient function. This method was introduced in [ 33 , 34 ] and later generalized in [ 37 , 7 ] . In 2008, Yau and Yau developed the Yau-Yau algorithm, a new method for solving the ”pathwise-robust” DMZ equation in time-invariant systems [ 42 ] . It has been theoretically proven that the Yau-Yau algorithm can converge to the true solution, as long as the growth rate of the observation |h| is greater than that of the drift |f| . Later, Luo and Yau extended the Yau-Yau algorithm to time-varying cases [ 21 ] . For readers interested in the mathematical principles and computational techniques enabling the software to address high-dimensional nonlinear filtering problems, we direct them to the work of Yueh et al. [ 23 , 24 ] . These publications provide a detailed analysis of the underlying mechanisms that power the software’s functionality. In this article, leveraging the powerful capabilities of R and C++, we have crafted an advanced software package. Rooted in the YauYau algorithm, this tool elegantly addresses the complexities of nonlinear filtering problems with remarkable precision and efficiency. This package delivers sophisticated numerical techniques for filtering, while seamlessly integrating an engaging Shiny application designed for dynamic visualization and in-depth exploration. This software package is designed to be concise and user-friendly, enabling users to quickly get started and operate it with ease. 2 The YauYauAL package 2.1 Dependency The YauYauAL package (v0.1.0) is built and tested on R version 4.4.2, which can work properly on a standard laptop computer with R version 4.4.2 or higher installed. Also, it is recommended to install RStudio for a better view and interaction with the objects stored in the R environment. The following dependencies are required for data processing: Deriv (v4.1.6), RcppArmadillo (v1.4.2.3-1), RcppEigen (v0.3.4.0.2), and Matrix (v1.7-1). In addition, ggplot2 (v3.5.1), reshape2