On decoupling of functions of normal vectors
Let \(Y=(Y_1,\dotsc,Y_d)\) be a random Gaussian vector with \(\operatorname{E}(Y_i)=0\), \(\operatorname{E}(Y_i^2)=1\), \(i=1,\dotsc, d\), and a positive-definite covariance matrix \(R\). Let \(c^-\) and \(c^+\) be the smallest and largest eigenvalues of the covariance matrix \(R\). The authors prove that \[ c^-\sum_{i=1}^d||\phi_i(Y_i)||^2_2\leq ||\sum_{i=1}^d\phi_i(Y_i)||_2^2\leq c^+\sum_{i=1}^d||\phi_i(Y_i)||^2_2 \] for all measurable functions \(\phi_i:\mathbb{R} \rightarrow \mathbb{R}\) satisfying the condition \(\operatorname{E}[\phi_i(Y_i)]=0\), \(i=1,\dotsc,d\). In addition, it was shown that the constants \(c^-\) and \(c^+\) are the best possible for the inequality to hold for all measurable functions \(\phi_i:\mathbb{R} \rightarrow \mathbb{R}\) satisfying the condition \(\operatorname{E}[\phi_i(Y_i)]=0\), \(i=1,\dotsc,d\).
This page was built for publication: On decoupling of functions of normal vectors
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1929771)
