Local stability of the output least square parameter estimation technique

From MaRDI portal





We investigate the well posedness of the OLS output least square setting of parameter estimation problems: find \(\hat x\in C\) which minimizes \(\| \phi (x)-z\|^ 2\) over C, where x is the parameter, C the convex of admissible parameters, z the measurement and \(\phi\) the parameter \(\to\) output mapping. When \(\| \phi '(x)y\| \geq \alpha \| y\|\) with \(\alpha >0\), and C is small enough compared to the curvature of \(\phi\), we obtain the (Lipschitz) continuity of the mapping \(z\to \hat x\) on a neighborhood of \(\phi\) (C). As a practical application of this result, we get that for finite dimensional parameters, the OLS technique yields an estimated parameter converging towards the exact one whenever the above hypotheses hold. Two examples of application are given.











This page was built for publication: Local stability of the output least square parameter estimation technique

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3683972)