VTechWorks staff will be away for the Thanksgiving holiday beginning at noon on Wednesday, November 27, through Friday, November 29. We will resume normal operations on Monday, December 2. Thank you for your patience.

Browsing by Author "Hamdan, Mohammed A."

Now showing 1 - 2 of 2
  • Thumbnail Image
    A Monte Carlo study of the robustness of the standard deviation of the sample correlation coefficient to the assumption of normality
    Brooks, Camilla Anita (Virginia Tech, 1970-03-08)
    From the case studies presented, one could conclude that for large values of n the standard deviation of r, the usual estimator of the correlation coefficient, and its transform z are only negligibly affected by variation in skewness or variation in kurtosis, the effect being slightly greater for variation in kurtosis. When the variations are in both skewness and kurtosis, the standard deviation of r and of Z are more affected by non-normality, a few significantly so. In small samples (n=10, n=5) the standard deviations of r and,z are quite visibly larger for variations in skewness and variations in kurtosis. The effect is greater for the simultaneous variation of the two. However, all of the values fall within a 95% confidence interval. It would appear then that the increase in the standard deviation of rand z is due more to the natural rise of the standard deviation in small samples rather than to non-normality. Viewing the studies made in totality we may in final conclusion state that the effect of non-normality on the standard deviation of r for samples of any size is not significant enough for concern; i.e., from this Monte Carlo study we will state that the standard deviation of the sample correlation coefficient is robust to the assumption of normality.
  • Thumbnail Image
    Some stochastic integral and discrete equations of the volterra and fredholm types with applications
    Padgett, W. J. (Virginia Tech, 1971-01-04)
    Random or stochastic integral equations occur frequently in the mathematical description of random phenomena in engineering, physics, biology, and oceanography. The present study is concerned with random or stochastic integral equations of the Volterra type in the form x(t;w) = h(tiW) + fa k(t,T~w)f(T,x(Tjw»dT, t > 0, and of the Fredholm type in the form 00 x(tjw) = h(t:w) + fa ko(t,T;w)e(T,x(T;w»dT, t ~ 0, where w £ Q, the supporting set of a complete probability measure space (n,A,p). A random function x(t:w} is said to be a random solution of an equation such as those above if it satisfies the equation with probability one. It is also required that X(tiW) be a second order stochastic process. The purpose of this dissertation is to investigate the existence, uniqueness, and stochastic stability properties of a random solution of these Volterra and Fredholm stochastic integral equations using the "theory of admissibility" and probabilistic functional analysis. The techniques of successive approximations and stochastic approximation are employed to approximate the random solution of the stochastic Volterra integral equation, and the convergence of the approximations to the unique random solution in mean square and with probability one is proven. Problems in telephone traffic theory, hereditary mechanics, population growth, and stochastic control theory are formulated, and some of the results of the investigation are applied. Finally, a discrete version of the above random integral equations is given, and several theorems concerning the existence, uniqueness, and stochastic stability of a random solution of the discrete equation are proven. Approximation of the random solution of the discrete version is obtained, and its convergence to the random solution is studied. This work extends and generalizes the work done by C. P. Tsokos in Mathematical Systems Theory 3 {1969}, pages 222-231, and M. W. Anderson in his Ph.D. dissertation at the University of Tennessee, 1966, among others. Extensions:of this research to several areas of application are proposed.