Skip to main content
ARS Home » Research » Publications at this Location » Publication #59424
Title: CONVERGENCE PROBLEMS - CHAPTER 4

Author
item RUSLING, JAMES - UNIV OF CONNECTICUT
item Kumosinski, Thomas

Submitted to: Nonlinear Computer Modeling of Chemical and Biochemical Data
Publication Type: Book / Chapter
Publication Acceptance Date: 5/4/1995
Publication Date: N/A
Citation: N/A

Interpretive Summary: Book Chapter: interpretive summary is not required.

Technical Abstract: Correlation between parameters and other convergence problems are discussed in chapter 4. Problems of correlations, convergence and the propagation of errors are presented. Correlations and how to minimize them by recognizing the correlated parameters. Orthogonalization methods and Fourier deconvolution techniques are powerful methods which can be used in conjunction with nonlinear regression analysis to determine theoretical parameters from certain experimental data. The pitfalls in convergence due to the limits on parameters (keeping parameters constant in preliminary analyses) as well as error surfaces are discussed. The propagation of computing errors for this analysis is also presented.