Concentration inequalities for polynomials in \(\alpha\)-sub-exponential random variables (Q2042767)

Let $X_1,\dots,X_n$ be independent random variables and $f:\mathbb R^n\to \mathbb R$ be a measurable function. One of the main questions of probability theory has been finding a good function $h:[0,\infty]\to[0,1]$ such that \[ P(|f(X_1,\dots,X_n)|-Ef(X_1,\dots,X_n)|\geq t)\leq h(t), \] and the following concentration inequality, regarding the tail decay of the normal distribution is perhaps the most well-known one: If $X_1,\dots,X_n$ are independent $N(0,1)$ random variables, and $f(X_1,\dots,X_n)=n^{-1/2}\sum_{i=1}^n X_i$, then $f(X_1,\dots,X_n)$ is distributed as $N(0,1)$ and \[ P(|f(X_1,\dots,X_n)|-Ef(X_1,\dots,X_n)|\geq t)\leq 2e^{-t^2/2}. \] The present authors derive concentration inequalities for polynomials $f(X_1,\dots,X_n)$ in independent random variables $X_1,\dots,X_n$ with an $\alpha$-sub-exponential tail decay. A particularly interesting case is given by quadratic forms $f(X_1,\dots,X_n)=\langle X, AX \rangle$ for which they prove Hanson-Wright-type inequalities [\textit{D. L. Hanson} and \textit{F. T. Wright}, Ann. Math. Stat. 42, 1079--1083 (1971; Zbl 0216.22203)] with explicit dependence on various norms of the matrix $A$. A consequence of these inequalities is a two-level concentration inequality for quadratic forms in $\alpha$-sub-exponential random variables, such as quadratic Poisson chaos. They provide various applications of these inequalities. Among them are generalizations of some results proven by \textit{M. Rudelson} and \textit{R. Vershynin} [Electron. Commun. Probab. 18, Paper No. 82, 9 p. (2013; Zbl 1329.60056)] from sub-normal to $\alpha$-sub-exponential random variables, i.e., concentration of the Euclidean norm of the linear image of a random vector and concentration inequalities for the distance between a random vector and a fixed subspace.