Proof. If the Xi are replaced by -Xi, the hn(r) do not be changed. Therefore we can assume a>O. Let H,(z) be the distribution function of the random
On the Almost Sure Convergence of Floating-Point Mantissas and Benford's Law
โ Scribed by Peter Schatte
- Publisher
- John Wiley and Sons
- Year
- 1988
- Tongue
- English
- Weight
- 238 KB
- Volume
- 135
- Category
- Article
- ISSN
- 0025-584X
No coin nor oath required. For personal study only.
โฆ Synopsis
Let Y1, Y2, ... be a sequence of random variables 8nd let J i n be the floating-point mantissa of Y,,. Further let l f ~, ~) ( , ) denote the indicator function of the interval [I, x). If Yn/n-Z as.. wheteZ+O is 5 fnrther random variable, then the seqirence i~~,z#3fn) converges a s . to log x in the sense of -X--means and logarithmic means, respectively. The speed of convergence in this relations is estimated. -4s a conclnsion. a further argument for BEXFORD'S law is provided.
๐ SIMILAR VOLUMES
Proof. If the Xi are replaced by -Xi, the hn(r) do not be changed. Therefore we can assume a>O. Let H,(z) be the distribution function of the random
In this paper, we consider the rate of convergence of the parameter estimation error and the cost function for the stochastic gradient-type algorithm. The problem is solved in the case of the minimum-variance stochastic adaptive control. It is proven that the cost function has the rate of convergenc