DIGITAL SIMULATION OF THE WIENER CANONICAL EXPANSION.
INFORMATION RESEARCH ASSOCIATES INC LEXINGTON MASS
Pagination or Media Count:
The numerical and programming aspects of a digital simulation of the Hermite-Laguerre Wiener-Canonical expansion of the likelihood function for a stochastic process are described as a first step in a computer investigation and evaluation of the Brick-Zames method of synthesizing Bayes Optimum Filters. Initial simulations are being conducted for the base-line cases of white-normal-noise and signal-plus-white-normal noise inputs. Present trial inputs are generated by Monte-Carlo methods. Numerical considerations governing the input process, the convergence of each Laguerre term, the processing time interval required for accurate approximation of the input process, convergence of the coefficients, and the Hermite polynomials are investigated. Where found necessary to investigate these, relevant computer diagnostic procedures are described. Initial results for the base-line processes, although preliminary, are discussed qualitatively. The next steps to be taken in the evaluation are also outlined. The computer programs, flow charts, and Fortran program listings are given in an Appendix. Various allowable forms of the Hermite-Laguerre expansion, the convergence properties of each, the relevant amplitude-scaling factors, and their relative advantages and disadvantages are considered in another Appendix. Author
- Computer Programming and Software