The Local Haar Condition in Parameter Estimation for Second Order Ordinary

Abstract

The problem considered is the computation of parameters in mathematical models described by scalar second order ordinary diierential equations of initial value type. The parameters are to be estimated from given data. The problem is treated as a nonlinear Chebyshev or least squares tting problem on a discrete set. Suu-cient conditions are obtained for the problems to satisfy the local Haar condition. This condition has important implications for the nature of best ts and can improve the performance of certain numerical methods. More importantly, it enables a local minimum to be recognised as a non{global minimum. Ilustrative numerical examples are described.

Cite this paper

@inproceedings{Kalogiratou1999TheLH, title={The Local Haar Condition in Parameter Estimation for Second Order Ordinary}, author={Z. Kalogiratou and Jack Williams}, booktitle={PDPTA}, year={1999} }