Mathematics
Power estimates of cuts of some improper integrals
A. V. Pozhidaev,
N. M. Pekel'nik,
O. I. Khaustova,
I. A. Trefilova Siberian Transport University, Novosibirsk
Abstract:
Background. Gaussian distribution arises naturally in many applications and is widely used in a variety of theoretical constructs. The important role is played by a lower cut-off function
$Q(x)$ of an improper integral from the density of a standard Gaussian distribution. The purpose of this paper is to obtain upper cuts for the arbitrary power of the function
$Q(x)$ through the improper integral of the same type with a lower limit
$ax$, where
$a$ - an arbitrary parameter.
Materials and methods. To obtain the necessary estimates the authors studied the behavior of the difference
$Q^m(x)-Q(\sqrt{m}x)$ in various intervals of the real axis. At the same time, the well-known properties of the Gaussian distribution were widely used. In addition, the strict inequalities were brought to a special form of the exponential function, and upper and lower bounds for the function
$Q(x)$ were obtained.
Results. The paper shows that for any real
$x$, when
$m>2$, the inequality
$Q^m(x)<Q(ax)$, where
$a$ - an arbitrary number in the interval
$[1;\sqrt{m}]$. In addition, it was found that this inequality cannot be improved on the parameter
$a$. So, the paper shows, that the right border of the interval for
$a$ can not be more than
$\sqrt{m}$ and the left - can not be less than 1.
Conclusions. The arbitrary degree function
$Q(x)$ can be uniformly bounded above by a function of the same type with
$ax$ argument. These estimates can be used in sociological and demographic studies in econometrics and statistics for point and interval estimates of the unknown parameters of the distribution.
Keywords:
probability density, gamma function, additional function of errors, logarithmically concave function, unimprovable values, Gaussian distribution, power estimations, distribution function.
UDC:
517