scientific article; zbMATH DE number 176891

[For the entire collection see Zbl 0755.00022.] This paper is concerned with the normalized area \(\omega_ n\) below a Bernoulli excursion \(\eta^ +_ 1,\dots,\eta^ +_{2n}\), defined by \(2n\omega_ n=\sum_ 1^{2n}\eta^ +_ i\) where \(\eta^ +_ 1,\dots,\eta^ +_{2n}\) is a Bernoulli random walk conditioned upon \(\eta^ +_ 0=\eta^ +_{2n}=0\) and \(\eta^ +_ i>0\) for \(0<i<2n\). The moments \(E\omega^ r_ n\) are explicitly found in terms of certain polynomials \(Q_ r(n)\), and constants \(M_ r\) are identified such that asymptotically \(E(\omega_ n/\sqrt{2n})^ r\approx M_ r\). The main result of the paper is an infinite series expansion for the limiting c.d.f. \(W(x)\) of \(\omega_ n/\sqrt{2n}\); the terms involve confluent hypergeometric functions and zeroes of the Airy function. Noting that \(W(x)\) is also the c.d.f. of a Brownian excursion, this improves upon results of Darling and Louchard. The proofs use combinatorial arguments, transforms and contour integration. A similar analysis is performed for the normalized area \(\sigma_ n\) in a tied-down random walk \(\eta_ 1,\dots,\eta_{2n}\), defined by \(2n\sigma_ n=\sum_ 1^{2n}|\eta_ i|\) where \(\eta_ 1,\dots,\eta_{2n}\) is a Bernoulli random walk conditioned upon \(\eta_ 0=\eta_{2n}=0\). The results are applied to determine the asymptotical distribution of the height of a random tree and of the statistic \(\int^ \infty_{- \infty}| F_ n(x)-G_ n(x)| dV(x)\) measuring the distance between the empirical c.d.f.s \(F_ n\), \(G_ n\) of two independent samples from the same underlying distribution \(V\).

Recommendations

Cited in

(13)

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 Q4036723)