Asymptotic analysis via Mellin transforms for small deviations in \(L^2\)-norm of integrated Brownian sheets (Q1884727)

The authors use Mellin transforms to compute a full asymptotic expansion for the tail of the Laplace transform of the squared \(L^2\)-norm of any multiply-integrated Brownian sheet. Through reversion they also obtain corresponding strong small-deviation estimates. Although Mellin transforms enjoy common use in analysis of algorithms [\textit{P. Flajolet} and \textit{R. Sedgewick}, ``Analytic combinatorics'' (book in preparation)], their application in this paper is the first one in studying small or large deviations. Their use is very natural and makes many computations completely transparent. The metods developed in this paper can be applied to many other Gaussian random fields besides multiply-integrated Brownian sheet. Consider a Gaussian random field whose covariance operator is the tensor product of marginal operators. If the Gaussian processes corresponding to these marginal operators each have a log Laplace transform whose asymptotic behavior can be established, then the methods of this paper will apply. Further, since one can derive a complete asymptotic expansion for the log Laplace transform, one needs only to produce a complete asymptotic expansion extending Sytaya's theorem in order to obtain a complete asymptotic expansion -- not just the lead-order asymptotics -- for the small-ball probability [\textit{G. N. Sytaya}, in: Teor. sluchajn. Protsess., respubl. mezhvedomst. Sb. 2, 93--104 (1974; Zbl 0282.60022)]. The authors mention that they will produce the needed generalization of Sytaya's theorem and apply it to small-ball probabilities for various Gaussian random fields in their future work.