Estimation and testing for partially functional linear errors-in-variables models (Q1733291): Difference between revisions

The following partially functional linear model it is considered \[ Y=\mathbb{Z}^T\beta+\int_{\mathcal{I}}\alpha(t)X(t)\text{d}t+\varepsilon, \] where $Y$ is a scalar response variable, $\mathbb{Z}$ is a $p$-dimensional vector of scalar covariates and $X$ is a smooth and square integrable random function defined on a bounded closed interval $\mathcal{I}$. It is assumed that the regression parameter $\alpha$ is also smooth and square integrable on $\mathcal{I}$. The random error $\varepsilon$ with zero mean and finite variance $\sigma^2$ is independent of $(\mathbb{Z}, X)$. The authors of the article propose estimators for the $p$-dimensional vector $\beta$ and for the unknown regression parameter $\alpha$ in the case when the vector $\mathbb{Z}$ is measured with error. It is supposed that instead of observing $\mathbb{Z}$ we observe the vector $\mathbb{W}$ such that $\mathbb{W}=\mathbb{Z}+\mathbb{U}$, where the $p$-dimensional error $\mathbb{U}$ with zero mean is independent of $(Y,\mathbb{Z},X)$. In the paper, the authors focus on the case where the covariance matrix of $\mathbb{U}$ is known.