Tail and moment estimates for chaoses generated by symmetric random variables with logarithmically concave tails (Q1930657)

A homogeneous polynomial chaos of order \(d\) is a random variable defined as \(\sum_{i_1,\dotsc,i_d=1}^na_{i_1,\dotsc,i_d}X_{i_1}\dots X_{i_d}\) where \(X_1,\dotsc,X_d\) is a sequence of independent real-valued random variables and \((a_{i_1,\dotsc,i_d})_{1\leq i_1,\dotsc,i_d\leq n}\) is a \(d\)-indexed symmetric array of real numbers satisfying \(a_{i_1,\dotsc,i_d}=0\) whenever there exists \(k \neq l\) such that \(i_k = i_l\). The authors derive two-sided bounds of moments and tails of polynomial chaoses of order at most three generated by independent symmetric random variables with log-concave tails. They also obtain such estimates for chaoses of arbitrary order generated by independent symmetric exponential variables. The bounds obtained are optimal up to constants depending only on the order \(d\) of the chaos variable.