A Max-Plus-Based Algorithm for a Hamilton--Jacobi--Bellman Equation of Nonlinear Filtering (Q4943741)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 1416967
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | A Max-Plus-Based Algorithm for a Hamilton--Jacobi--Bellman Equation of Nonlinear Filtering |
scientific article; zbMATH DE number 1416967 |
Statements
A Max-Plus-Based Algorithm for a Hamilton--Jacobi--Bellman Equation of Nonlinear Filtering (English)
0 references
19 March 2000
0 references
max-plus algebra
0 references
nonlinear HJB equations
0 references
robust filtering
0 references
0.8898411
0 references
0.8799741
0 references
0.8620762
0 references
0.86040324
0 references
0.8574998
0 references
0.8546376
0 references
0.85388756
0 references
0 references
0.85095906
0 references
This paper is concerned with an algorithm for construction of viscosity solutions of a certain Hamilton-Jacobi-Bellman (HJB) equation. The solution of this partial differential equation (PDE) is a necessary step in the application of certain filtering techniques (such as robust filtering) to nonlinear systems. The HJB equation for the filter information state is a first-order equation with a term that is quadratic in the gradient. However, in the max-plus algebra, the solution operator is linear. The algorithm takes advantage of this linearity by using a max-plus basis representation of the solution. Propagation of the solution is reduced to a max-plus matrix multiplication on the coefficients of the basis representation. Section 2 of the paper reviews the robust \(H_\infty\) filter (and Mortensen filter) and provides motivation for the solution of the relevant HJB equations. In the next section, the max-plus algebra is described, and the max-plus linearity of the solution operator is proved. A semiconvex duality-based representation of the solution of the PDE is included in section 4. In section 5, the semiconvex duality representation is used to obtain the basis representation. Both the case of discrete-time measurements and the case of continuous-time measurements are considered. In the second case, a proof of convergence of the algorithm as the measurement time-step goes to zero is presented in the last section.
0 references