The limit behavior of an interval splitting scheme (Q1080269)

The interval [0,1] is split by generating a uniformly distributed random variable on that interval. The largest of the two intervals thus obtained is split again, and this process is continued ad infinitum. The authors show that the extremes of this intervals converge almost surely to a random variable W. The limit variable W exists because the extremes are both monotone sequences. The random variables W and \(Z=WU\), if \(U>\), and \(=1-W(1-U)\), if \(U\leq\) have the same distribution where U is a uniformly distributed random variable on [0,1] and independent of W. It follows from this that the distribution of W is the beta(2,2).