GENERAL PROBLEMS OF METROLOGY AND MEASUREMENT TECHNIQUES
Z. I. Sharagina, S. K. Ivanov
Abstract
(i) is therefore a good reflection of the crux of the measurement process. Here H(X) is the which characterizes the a priori of the quantity being measured, and H(X/Y) is the total conditional which characterizes the of the quantity being measured which remains after the measurement result Y has been obtained. Let us examine the application of the concept in the case of continuous distributions. The value of the as a numerical characteristic of a continuous random variable is rejected in [12] on the basis that the equals ~ for continuous random variables. The introduction of the probability density p(x) could be rejected on approximately the same basis by considering that this quantity does not reflect the probabilistic characteristic by not being a probability; and only the product p(x)dx possesses the probabilistic interpretation, but equals 0 for continuous quantities and hence loses meaning. Wherever required and possible, the probability is approximated by the expression p(x)Ax, where &x, albeit small, is a finite quantity. The concepts of reduced (differential) entropy and quantization step have an analogous (but not identical) meaning [16]. If is understood to be the mathematical expectation of the logarithm of the probability of random variable values, then it is indeed impossible to relate this concept to continuous random variables. On the other hand, the initial physical meaning, uncertainty, vanishes in the reduced entropy concept. The initial concept uncertainty not only remains real for continuous random variables in the examination of the quantization step concept, but stands out more, exposing most completely the content of the Heisenberg principle and the negative principle of information [5, 7], according to which no quantity being observed can be measured with arbitrarily high accuracy. The limit is not set on this accuracy by the imperfection of the instruments but by objective laws of nature.