scientific article; zbMATH DE number 1533546
Exponential bounds for the distribution of norms of sub-Gaussian random vectors taking its values in \(R^n\) are found. The random vector \(X\) taking its values in \(R^n\) is called sub-Gaussian with respect to a symmetric positive definite \(n \times n\) matrix \(R\) if for all \(x \in R^n\) the following inequality holds true \[ E \exp\{ (x, X)\} \leq \exp\bigl\{\tfrac{1}{2}(Rx,x)\bigr\}. \] Let \(k\) be an integer, \(0\leq k\leq 1\). Denote by \(I\) the set \(I =\{ i_1, \ldots,i_k\}\) for \(k \geq 1\), \(I = \emptyset\) for \(k=0\), where \(i_1, \ldots,i_k\) are integers such that \(1 \leq i_1<i_2\ldots<i_k\leq n\). Let \(M_I =\{ m_{ij}^{(I)}\}\) be a matrix with \(m_{jj}^{(I)} =1\) for \(i=j \in I\), \(m_{ij}^{(I)} = -1\) for \(i=j \in I\), \(m_{ij}^{(I)} =0\) elsewhere. Let \(R =\{ r_{ij}\}\) be any \(n \times n\) matrix \(R_I =M_I R M_I\). Let \(\xi^{(n)}\) be the vector in \(R^n\) whose components are all \(1\). \(R\) has property P iff for every \(I\) the vector \(R_I^{-1} \xi^{(n)}\) has all strictly positive components and the same happens for each of the diagonal sub-matrices of \(R.\) The following theorem is the main result of this paper: Let \(X =(X_1,\ldots, X_n)^\top\) be sub-Gaussian with respect to \(R\) and assume that \(R\) has property P, then for every \(t>0\) we have \[ P\bigl\{ \|X\|> t\bigr\} \leq {3^n +(-1)^{n-1} \over 2} \exp \biggl\{ -{t^2 \over 2n}\alpha \biggr\}, \] where \(\alpha\) is a number depending only on \(R\). \(\alpha\) is identified in this paper.
- Exponential integrability of sub-Gaussian vectors
- Norms of sub-exponential random vectors
- The sub-Gaussian norm of a binary random variable
- scientific article; zbMATH DE number 4020117 (Why is no real title available?)
- scientific article; zbMATH DE number 1740315 (Why is no real title available?)
- scientific article; zbMATH DE number 1437709 (Why is no real title available?)
- On strict sub-Gaussianity, optimal proxy variance and symmetry for bounded random variables
- Subgaussianity is hereditarily determined
- Research and application of Hoeffding inequality under sub-Gaussian distribution
- Super-Gaussian directions of random vectors
- A simple tool for bounding the deviation of random matrices on geometric sets
- Inequalities for the distributions of functionals of sub-Gaussian vectors
- A note on the asymptotic behavior of sequences of generalized subgaussian random vectors
- Matrix deviation inequality for ℓp-norm
This page was built for publication:
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4516951)