Showing 1–8 of 8 results for author: Székely, G J

Three classical probability problems: the Hungarian roulette

Authors: Tamás F. Móri, Gábor J. Székely

Abstract: In this paper three unrelated problems will be discussed. What connects them is the rich methodology of classical probability theory. In the first two problems we have a complete answer to the problems raised; in the third case, what we call the Hungarian roulette problem, we only have a conjecture with heuristic justification. In this paper three unrelated problems will be discussed. What connects them is the rich methodology of classical probability theory. In the first two problems we have a complete answer to the problems raised; in the third case, what we call the Hungarian roulette problem, we only have a conjecture with heuristic justification. △ Less

Submitted 27 June, 2024; v1 submitted 16 April, 2024; originally announced April 2024.

Comments: 13 pages, 1 figure. To appear in Annales Univ. Sci. Budapest., Sec. Comp., vol. 56

MSC Class: 60E10; 60F05; 62E10

The Earth Mover's Correlation

Authors: Tamás F. Móri, Gábor J. Székely

Abstract: Since Pearson's correlation was introduced at the end of the 19th century many dependence measures have appeared in the literature. Recently we have suggested four simple axioms for dependence measures of random variables that take values in Hilbert spaces. We showed that distance correlation satisfies all these axioms. We still need a new measure of dependence because existing measures either do… ▽ More Since Pearson's correlation was introduced at the end of the 19th century many dependence measures have appeared in the literature. Recently we have suggested four simple axioms for dependence measures of random variables that take values in Hilbert spaces. We showed that distance correlation satisfies all these axioms. We still need a new measure of dependence because existing measures either do not work in general metric spaces (that are not Hilbert spaces) or they do not satisfy our four simple axioms. The earth mover's correlation introduced in this paper applies in general metric spaces and satisfies our four axioms (two of them in a weaker form). △ Less

Submitted 9 September, 2020; originally announced September 2020.

MSC Class: 62H20

Calibrating dependence between random elements

Authors: Abram M. Kagan, Gabor J. Székely

Abstract: Attempts to quantify dependence between random elements X and Y via maximal correlation go back to Gebelein (1941) and Rényi (1959). After summarizing properties (including some new) of the Rényi measure of dependence, a calibrated scale of dependence is introduced. It is based on the ``complexity`` of approximating functions of X by functions of Y. Attempts to quantify dependence between random elements X and Y via maximal correlation go back to Gebelein (1941) and Rényi (1959). After summarizing properties (including some new) of the Rényi measure of dependence, a calibrated scale of dependence is introduced. It is based on the ``complexity`` of approximating functions of X by functions of Y. △ Less

Submitted 11 March, 2019; originally announced March 2019.

MSC Class: 60H99; 62E10

Schur properties of convolutions of gamma random variables

Authors: Farbod Roosta-Khorasani, Gabor J. Szekely

Abstract: Sufficient conditions for comparing the convolutions of heterogeneous gamma random variables in terms of the usual stochastic order are established. Such comparisons are characterized by the Schur convexity properties of the cumulative distribution function of the convolutions. Some examples of the practical applications of our results are given. Sufficient conditions for comparing the convolutions of heterogeneous gamma random variables in terms of the usual stochastic order are established. Such comparisons are characterized by the Schur convexity properties of the cumulative distribution function of the convolutions. Some examples of the practical applications of our results are given. △ Less

Submitted 18 January, 2016; originally announced January 2016.

MSC Class: 60E15; 62E99

Journal ref: Metrika 78(8):997-1014, 2015

Assessing stochastic algorithms for large scale nonlinear least squares problems using extremal probabilities of linear combinations of gamma random variables

Authors: Farbod Roosta-Khorasani, Gábor J. Székely, Uri Ascher

Abstract: This article considers stochastic algorithms for efficiently solving a class of large scale non-linear least squares (NLS) problems which frequently arise in applications. We propose eight variants of a practical randomized algorithm where the uncertainties in the major stochastic steps are quantified. Such stochastic steps involve approximating the NLS objective function using Monte-Carlo methods… ▽ More This article considers stochastic algorithms for efficiently solving a class of large scale non-linear least squares (NLS) problems which frequently arise in applications. We propose eight variants of a practical randomized algorithm where the uncertainties in the major stochastic steps are quantified. Such stochastic steps involve approximating the NLS objective function using Monte-Carlo methods, and this is equivalent to the estimation of the trace of corresponding symmetric positive semi-definite (SPSD) matrices. For the latter, we prove tight necessary and sufficient conditions on the sample size (which translates to cost) to satisfy the prescribed probabilistic accuracy. We show that these conditions are practically computable and yield small sample sizes. They are then incorporated in our stochastic algorithm to quantify the uncertainty in each randomized step. The bounds we use are applications of more general results regarding extremal tail probabilities of linear combinations of gamma distributed random variables. We derive and prove new results concerning the maximal and minimal tail probabilities of such linear combinations, which can be considered independently of the rest of this paper. △ Less

Submitted 27 November, 2014; v1 submitted 1 April, 2014; originally announced April 2014.

MSC Class: 65C20; 65C05; 60E05; 68W20

Journal ref: SIAM/ASA Journal on Uncertainty Quantification. 3 (2015) 61-90

Partial Distance Correlation with Methods for Dissimilarities

Authors: Gabor J. Szekely, Maria L. Rizzo

Abstract: Distance covariance and distance correlation are scalar coefficients that characterize independence of random vectors in arbitrary dimension. Properties, extensions, and applications of distance correlation have been discussed in the recent literature, but the problem of defining the partial distance correlation has remained an open question of considerable interest. The problem of partial distanc… ▽ More Distance covariance and distance correlation are scalar coefficients that characterize independence of random vectors in arbitrary dimension. Properties, extensions, and applications of distance correlation have been discussed in the recent literature, but the problem of defining the partial distance correlation has remained an open question of considerable interest. The problem of partial distance correlation is more complex than partial correlation partly because the squared distance covariance is not an inner product in the usual linear space. For the definition of partial distance correlation we introduce a new Hilbert space where the squared distance covariance is the inner product. We define the partial distance correlation statistics with the help of this Hilbert space, and develop and implement a test for zero partial distance correlation. Our intermediate results provide an unbiased estimator of squared distance covariance, and a neat solution to the problem of distance correlation for dissimilarities rather than distances. △ Less

Submitted 8 July, 2014; v1 submitted 10 October, 2013; originally announced October 2013.

MSC Class: Primary: 62H20; 62H15; Secondary: 62Gxx

Measuring and testing dependence by correlation of distances

Authors: Gábor J. Székely, Maria L. Rizzo, Nail K. Bakirov

Abstract: Distance correlation is a new measure of dependence between random vectors. Distance covariance and distance correlation are analogous to product-moment covariance and correlation, but unlike the classical definition of correlation, distance correlation is zero only if the random vectors are independent. The empirical distance dependence measures are based on certain Euclidean distances between… ▽ More Distance correlation is a new measure of dependence between random vectors. Distance covariance and distance correlation are analogous to product-moment covariance and correlation, but unlike the classical definition of correlation, distance correlation is zero only if the random vectors are independent. The empirical distance dependence measures are based on certain Euclidean distances between sample elements rather than sample moments, yet have a compact representation analogous to the classical covariance and correlation. Asymptotic properties and applications in testing independence are discussed. Implementation of the test and Monte Carlo results are also presented. △ Less

Submitted 28 March, 2008; originally announced March 2008.

Comments: Published in at http://dx.doi.org/10.1214/009053607000000505 the Annals of Statistics (http://www.imstat.org/aos/) by the Institute of Mathematical Statistics (http://www.imstat.org)

Report number: IMS-AOS-AOS0283 MSC Class: 62G10 (Primary) 62H20 (Secondary)

Journal ref: Annals of Statistics 2007, Vol. 35, No. 6, 2769-2794

Student's $t$-test for scale mixture errors

Authors: Gábor J. Székely

Abstract: Generalized t-tests are constructed under weaker than normal conditions. In the first part of this paper we assume only the symmetry (around zero) of the error distribution (i). In the second part we assume that the error distribution is a Gaussian scale mixture (ii). The optimal (smallest) critical values can be computed from generalizations of Student's cumulative distribution function (cdf),… ▽ More Generalized t-tests are constructed under weaker than normal conditions. In the first part of this paper we assume only the symmetry (around zero) of the error distribution (i). In the second part we assume that the error distribution is a Gaussian scale mixture (ii). The optimal (smallest) critical values can be computed from generalizations of Student's cumulative distribution function (cdf), $t_n(x)$. The cdf's of the generalized $t$-test statistics are denoted by (i) $t_n^S(x)$ and (ii) $t_n^G(x)$, resp. As the sample size $n\to \infty$ we get the counterparts of the standard normal cdf $Φ(x)$: (i) $Φ^S(x):=\operatorname {lim}_{n\to \infty}t_n^S(x)$, and (ii) $Φ^G(x):=\operatorname {lim}_{n\to \infty}t_n^G(x)$. Explicit formulae are given for the underlying new cdf's. For example $Φ^G(x)=Φ(x)$ iff $|x|\ge \sqrt{3}$. Thus the classical 95% confidence interval for the unknown expected value of Gaussian distributions covers the center of symmetry with at least 95% probability for Gaussian scale mixture distributions. On the other hand, the 90% quantile of $Φ^G$ is $4\sqrt{3}/5=1.385... >Φ^{-1}(0.9)=1.282...$. △ Less

Submitted 27 October, 2006; originally announced October 2006.

Comments: Published at http://dx.doi.org/10.1214/074921706000000365 in the IMS Lecture Notes--Monograph Series (http://www.imstat.org/publications/lecnotes.htm) by the Institute of Mathematical Statistics (http://www.imstat.org)

Report number: IMS-LNMS49-LNMS4902 MSC Class: 62F03 (Primary) 62F04 (Secondary)

Journal ref: IMS Lecture Notes--Monograph Series 2006, Vol. 49, 9-15