Showing 1–6 of 6 results for author: Volkova, K

Efficiency of Exponentiality Tests Based on a Special Property of Exponential Distribution

Authors: Ya. Yu. Nikitin, K. Yu. Volkova

Abstract: New goodness-of-fit tests for exponentiality based on a particular property of exponential law are constructed. Test statistics are functionals of U-empirical processes. The first of these statistics is of integral type, the second one is a Kolmogorov type statistic. We show that the kernels corresponding to our statistics are nondegenerate. The limiting distributions and large deviations of new s… ▽ More New goodness-of-fit tests for exponentiality based on a particular property of exponential law are constructed. Test statistics are functionals of U-empirical processes. The first of these statistics is of integral type, the second one is a Kolmogorov type statistic. We show that the kernels corresponding to our statistics are nondegenerate. The limiting distributions and large deviations of new statistics under the null hypothesis are described. Their local Bahadur efficiency for various parametric alternatives is calculated and is compared with simulated powers of new tests. Conditions of local optimality of new statistics in Bahadur sense are discussed and examples of "most favorable" alternatives are given. New tests are applied to reject the hypothesis of exponentiality for the length of reigns of Roman emperors which was intensively discussed in recent years. △ Less

Submitted 7 March, 2016; originally announced March 2016.

Comments: 19 pages, 5 figures

MSC Class: 60F10; 62F03; 62G20; 62G30

Bifractional Brownian motion: existence and border cases

Authors: Mikhail Lifshits, Ksenia Volkova

Abstract: Bifractional Brownian motion (bfBm) is a centered Gaussian process with covariance \[ R^{(H,K)}(s,t)= 2^{-K} \left( \left(|s|^{2H}+|t|^{2H} \right)^{K}-|t-s|^{2HK}\right), \qquad s,t\in R. \] We study the existence of bfBm for a given pair of parameters $(H,K)$ and encounter some related limiting processes. Bifractional Brownian motion (bfBm) is a centered Gaussian process with covariance \[ R^{(H,K)}(s,t)= 2^{-K} \left( \left(|s|^{2H}+|t|^{2H} \right)^{K}-|t-s|^{2HK}\right), \qquad s,t\in R. \] We study the existence of bfBm for a given pair of parameters $(H,K)$ and encounter some related limiting processes. △ Less

Submitted 5 November, 2015; v1 submitted 8 February, 2015; originally announced February 2015.

MSC Class: 60G15; 42A82

Journal ref: ESAIM PS, 2015, 19, 766--781

On asymptotic efficiency of goodness-of-fit tests for the Pareto distribution based on its characterization

Authors: K. Yu. Volkova

Abstract: We introduce a new characterization of Pareto distribution and construct integral and supremum type goodness-of-fit tests based on it. Limiting distribution and large deviations of new statistics are described and their local Bahadur efficiency for parametric alternatives is calculated. Conditions of local optimality of new statistics are given. We introduce a new characterization of Pareto distribution and construct integral and supremum type goodness-of-fit tests based on it. Limiting distribution and large deviations of new statistics are described and their local Bahadur efficiency for parametric alternatives is calculated. Conditions of local optimality of new statistics are given. △ Less

Submitted 20 August, 2014; originally announced August 2014.

Comments: 21 pages, 3 figures. arXiv admin note: text overlap with arXiv:1405.7210

MSC Class: 60F10; 62G10; 62G20; 62G30

Tests of exponentiality based on Arnold-Villasenor characterization, and their efficiencies

Authors: M. Jovanovic, B. Milosevic, Ya. Yu. Nikitin, M. Obradovic, K. Yu. Volkova

Abstract: We propose two families of scale-free exponentiality tests based on the recent characterization of exponentiality by Arnold and Villasenor. The test statistics are based on suitable functionals of U-empirical distribution functions. The family of integral statistics can be reduced to V- or U-statistics with relatively simple non-degenerate kernels. They are asymptotically normal and have reasonabl… ▽ More We propose two families of scale-free exponentiality tests based on the recent characterization of exponentiality by Arnold and Villasenor. The test statistics are based on suitable functionals of U-empirical distribution functions. The family of integral statistics can be reduced to V- or U-statistics with relatively simple non-degenerate kernels. They are asymptotically normal and have reasonably high local Bahadur efficiency under common alternatives. This efficiency is compared with simulated powers of new tests. On the other hand, the Kolmogorov type tests demonstrate very low local Bahadur efficiency and rather moderate power for common alternatives,and can hardly be recommended to practitioners. We also explore the conditions of local asymptotic optimality of new tests and describe for both families special "most favorable" alternatives for which the tests are fully efficient. △ Less

Submitted 18 July, 2014; originally announced July 2014.

Comments: 25 pages, 4 figures

MSC Class: 60F10; 62G10; 62G20; 62G30

Tests of exponentiality based on Yanev-Chakraborty characterization, and their efficiency

Authors: K. Y. Volkova

Abstract: We construct new tests of exponentiality based on Yanev-Chakraborty's characterization of exponential law. We calculate limiting distributions of new tests, local Bahadur efficiency for common alternatives and describe conditions of their local asymptotic optimality. We construct new tests of exponentiality based on Yanev-Chakraborty's characterization of exponential law. We calculate limiting distributions of new tests, local Bahadur efficiency for common alternatives and describe conditions of their local asymptotic optimality. △ Less

Submitted 28 May, 2014; originally announced May 2014.

Comments: 14 pages, 2 figures

Asymptotic Efficiency of Goodness-of-fit Tests for the Power Function Distribution Based on Puri--Rubin Characterization

Authors: Ya. Yu. Nikitin, K. Yu. Volkova

Abstract: We construct integral and supremum type goodness-of-fit tests for the family of power distribution functions. Test statistics are functionals of $U-$empirical processes and are based on the classical characterization of power function distribution family belonging to Puri and Rubin. We describe the logarithmic large deviation asymptotics of test statistics under null-hypothesis, and calculate thei… ▽ More We construct integral and supremum type goodness-of-fit tests for the family of power distribution functions. Test statistics are functionals of $U-$empirical processes and are based on the classical characterization of power function distribution family belonging to Puri and Rubin. We describe the logarithmic large deviation asymptotics of test statistics under null-hypothesis, and calculate their local Bahadur efficiency under common parametric alternatives. Conditions of local optimality of new statistics are given. △ Less

Submitted 3 July, 2012; originally announced July 2012.

Comments: 11 pages

MSC Class: 62F03; 62G20; 60F10