Geometric properties of monotone functions and probabilities of random fluctuations (Q1358090)

The chord joining two points of the graph of a real function becomes a rather interesting object of study as one of its endpoints starts sliding along the graph while the other remains fixed. Observing the variation of its slope, we can notice curious regularities that sometimes fall beyond the scope of the theory of functions of a real variable and reveal themselves in the properties of various processes in ergodic theory or probability. A classical example here is the celebrated Riesz lemma underlying his proof of Lebesgue's differentiation theorem and known as the maximal ergodic theorem in measure theory. We study the behavior of a chord of the graph of a monotone function. The fluctuations within a preassigned range are discussed for the slope of a chord. We establish some exact inequalities that show how often one can meet with a chord making a given number of such fluctuations. The meaning of these inequalities is very simple: sometime a chord may change its slope rapidly or even instantly in the direction corresponding to the character of monotonicity of the function under study, but afterwards it is always difficult for the chord to return the slope to the former level. The thrilling competition between the slope of a chord and the height of its moving endpoint results in the fact that the probability of fluctuations vanishes at the rate of a geometric progression as the number of fluctuations increases; moreover, the convergence is uniform in the whole class of monotone functions.