No label defined (Q3980259)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 21229
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | No label defined |
scientific article; zbMATH DE number 21229 |
Statements
26 June 1992
0 references
weighted metrics
0 references
Gaussian processes
0 references
order statistics
0 references
empirical quantile function
0 references
uniform quantile process
0 references
rate of convergence
0 references
Gaussian approximation
0 references
weighted quantile processes
0 references
weighted supremum norm behaviour
0 references
exact order in probability of weighted differences
0 references
0.8779810667037964
0 references
0.8728777170181274
0 references
0.840075671672821
0 references
The present paper is concerned with the asymptotic distributions of the suprema of weighted uniform and general quantile processes. For this, given i.i.d. r.v. \(X_ 1,X_ 2,\dots\) with an absolutely continuous distribution function \(F\), having a positive density \(f\) on the open support of \(F\), let \(Q\) be the inverse of \(F\), \(X_{1,n}\leq X_{2,n}\leq\dots\leq X_{n,n}\) be the order statistics of \(X_ i\), \(1\leq i\leq n\), and define the empirical quantile function \(Q_ n\) by \[ Q_ n(s):=X_{k,n},\;\;(k-1)/n<s\leq k/n,\;\;1\leq k\leq n. \] The quantile process \(\rho_ n\) is defined by \[ \rho_ n(s):=n^{1/2}f(Q(s))(Q(s)-Q_ n(s)),\;\;0<s<1. \] Put \(U_ i:=F(X_ i)\) to obtain i.i.d. \(U(0,1)\) r.v.'s with their corresponding order statistics \(U_{1,n}\leq U_{2,n}\leq\dots\leq U_{n,n}\) and empirical quantile functions \[ U_ n(s):=U_{k,n},\;\;(k-1)/n<s\leq k/n,\;\;1\leq k\leq n. \] The uniform quantile process \(u_ n\) is then defined by \[ u_ n(s):=n^{1/2}(s-U_ n(s)),\;\;0<s\leq 1. \] It is known from \textit{M. Csörgö} and \textit{P. Révész} [Ann. Stat. 6, 882-894 (1978; Zbl 0378.62050)] that under appropriate regularity conditions on \(F\) and \(f\) one obtains a fast enough rate of convergence to zero for \[ \sup_{1/(n+1)\leq s\leq n/(n+1)}|\rho_ n(s)-u_ n(s)| \] so that appropriate Gaussian approximations for \(u_ n\) became also valid for \(\rho_ n\). As to the asymptotic distributions of weighted quantile processes, \textit{M. Csörgö}'s monograph, Quantile processes with statistical applications. (1983; Zbl 0518.62043), is an excellent source of interesting results. In the present paper the authors point out that the weighted supremum norm behaviour of \(u_ n\) and that of \(\rho_ n\) can be the same or it can also be different, depending on the weights used and also on the part of the unit interval where the supremum is taken. Also, the exact order in probability of weighted differences of \(\rho_ n\) and \(u_ n\) is determined.
0 references