The study of probability distributions of the hyperbolic cosine type

M. S. Tokmachev

Vestnik Ûžno-Uralʹskogo gosudarstvennogo universiteta. Seriâ, Matematika, mehanika, fizika, Tome 9 (2017) no. 3, pp. 18-26

Résumé

Probability distribution, obtained earlier by the author in a result of the characterization of distributions by the property of the constant regression of quadratic statistics on a linear form and called the hyperbolic cosine type distribution, is considered. Along with the classic normal distribution, gamma distribution, negative binomial and some other distributions, the given three-parameter distribution is related to the class of probability distributions, united by a common characterization condition. The obtained distribution is a generalization of the famous two-parameter Meixner distribution. In the article, a proof that a function appeared in the result of given characterization is indeed characteristic is given. The structure of this function, in connection with gamma distribution and the corresponding distribution conjugated with the gamma distribution, along with the constant distribution, is presented. Infinite decomposability of distribution is proved. Based on characteristic function, the probability density function, expressed through the beta function of complex conjugate arguments is recovered in general terms. Along with a unified formula, correlations in elementary functions are deduced for the distribution density functions at integer values of parameter m. Density formulas at odd and even values of the parameter are similar on the structure of cofactors: exponent function, hyperbolic secant or cosecant correspondingly, and polynomial factors. The distribution under study has multiple applications, not only in probability problems. Moments of distribution at specified parameters change as polynomials of some class with corresponding coefficients. These coefficients can be constructed as number sets (number triangles and number sequences), both known and new with setting of a row of functional relationships.

Détail

Keywords: distribution of the hyperbolic cosine type, characteristic function, infinitely divisible distribution, beta function.

Bibliographie

[1] Tokmachev M. S., Dep. in VINITI 21.06.94, No 1542-B94, 11 pp. (in Russ.)

[2] Tokmachev M. S., “Constancy of regression of quadratic statistics with linear statistics”, Bulletin of the Novgorod state University, 1995, no. 1, 139–141 (in Russ.)

[3] Kagan A. M., Linnick Y. V., Rao S. R., Characterization problems of mathematical statistics, Nauka Publ., M., 1972, 656 pp. (in Russ.)

[4] Klebanov L. B., When the quadratic statistic has constant regression on the sample mean?, Theory of probability and its applications, XXIV:3 (1979), 646–648 (in Russ.) | Zbl

[5] C.D. Lai, “Meixner classes and Meixner hypergeometric distributions”, Australian New Zealand Journal of Statistics, 24 (1982), 221–233 | DOI | MR | Zbl

[6] C.D. Lai, “A characterization of gamma, Meixner hypergeometric and negative binomial distributions based on canonical measures”, Annals of the Institute of Statistical Mathematics, 34:1 (1982), 359–363 | DOI | MR | Zbl

[7] Vadzinskiy R. N., Handbook of Probabilistic Distributions, Nauka Publ., St. Petersburg, 2001, 295 pp. (in Russ.)

[8] Smirnov V. I., Course of Higher Mathematics, part 2, v. 3, Nauka Publ., M., 1974, 672 pp. (in Russ.)

[9] Lukacs E., Characteristic functions, Hafner Publishing Co., New York, 1970, 350 pp. | MR

[10] Tokmachev M. S., Tokmachev A. M., “Distribution type hyperbolic cosine”, Bulletin of the Novgorod state University, 2001, no. 17, 85–88 (in Russ.)

[11] Tokmachev M. S., “Applied aspect of the generalized hyperbolic cosine distribution”, Bulletin of the Novgorod state University, 2005, no. 34, 96–99 (in Russ.)

[12] Tokmachev M. S., “Some integrals associated with the distribution type hyperbolic cosine”, Mathematics in high school and in school, Proceedings of the XXIV International Scientific and Methodical Conference, Peterburgskiy gos. universitet putey soobshcheniya Publ., St. Petersburg, 2012, 185–186 (in Russ.)

[13] Tokmachev M. S., “Evaluation of integrals for some class of functions with probabilistic interpretation”, Bulletin of the Novgorod state University. Ser.: Physical and mathematical Sciences, 2014, no. 80, 42–46 (in Russ.)

[14] Tokmachev M. S., “About number sets and sequences in connection with a distribution of the hyperbolic cosine type”, Bulletin of the Novgorod state University. Ser.: Physical and mathematical Sciences, 2015, no. 3, 35–39 (in Russ.)

[15] Tokmachev M. S., Dep. in VINITI 09. 06. 2016, No 91-B2016, 90 pp. (in Russ.)