Contraction and decoupling inequalities for multilinear forms and \(U\)-statistics (Q1904479): Difference between revisions

The authors study contraction and decoupling inequalities for multilinear forms and \(U\)-statistics. The decoupling principles, for homogeneous random forms, are essential tools in multiple integration. These principles are shown usually under the symmetry hypothesis for the random variables. The authors prove decoupling inequalities for not necessarily symmetric random variables. They also stay and prove a decoupling principle by means of probability tails, getting that the two types of chaos (the decoupled and coupled) are simultaneously tight. The methods of the proofs are elementary and based in the algebraic nature of decoupling. They obtain several applications: Decoupling results for arbitrary rearrangement invariant norms and Orlicz functionals, a decoupling principle for \(U\)-statistics; they also show tail probability decoupling results.