Stieltjes integration and stochastic calculus with respect to self-affine functions (Q1179786)

A continuous function \(f:[0,1]\to R\) is a self-affine function of base \(\alpha\in(0,1)\) if its graph has a fractal-type property, \(\alpha\) being the parameter of the scaling transformation. The purpose of this paper is to present a sort of a deterministic stochastic calculus in the framework of self-affine functions of order \(\alpha=1/2\). Typically, such a function \(f\) has unbounded variation, in particular, the classical Stieltjes integral cannot be defined. However, the authors construct a so-called `order-two Stieltjes integral' of a function \(h(x)=g(x,f(x))\) (where \(g:R^ 2\to R\) is a \(C^ 2\) function) with respect to \(f\). They establish an analogue of the Itô formula (Itô's formula is the Fundamental Theorem of Stochastic Calculus). In the case when the image of the Lebesgue measure by \(f\) is absolutely continuous w.r.t. the Lebesgue measure, the authors obtain a Tanaka-like formula that provides an expression for the density (i.e. the local times) in terms of an order-two Stieltjes integral. A Stratonovich's formula is also proven, and the example when \(f\) is the so-called Rudin-Shapiro function is discussed. Previously, \textit{H. Föllmer} [Lect. Notes Math. 850, 143-150 (1981; Zbl 0461.60074)] and the reviewer [Ann. Probab. 17, No. 4, 1521- 1535 (1989; Zbl 0687.60054)] developed analogues of the stochastic calculus for deterministic functions, using a fixed sequence of partitions. Here, the authors use a second order averaging method, which gives more robustness.