Probabilistic \(L^p\) spaces (Q1947312)

Let \(D\) be the set of all distribution functions -- non-decreasing and left continuous functions \(F:\mathbb R \to [0,1]\) with \(\lim_{t \to -\infty} F(t)=0\) and \(\lim_{t \to +\infty} F(t)=1\) --, and let \(D^+\) be the set of those functions in \(D\) that satisfy \(F(0)=0\). Having in mind the notion of probabilistic normed spaces, where the usual norm in the ball of a normed space is substituted by a ``probabilistic norm function'' having values in \(D^+\), the authors define the notion of probabilistic (probability) measure by changing the target space of the function from the interval \([0,1]\) to \(D^+\). After Sections 2 and 3, where the authors develop some results regarding the structure of \(D\) and a new distance for this space, the notion of probabilistic valued measure and the integration with respect to it is defined and analysed. After obtaining some fundamental results on it, the corresponding integration is defined and studied in a similar fashion as it is done in vector measure integration for positive vector measures. The authors combine the properties of the space \(D^+\) together with an order-based integration structure (Definition 4.6 and Lemma 4.7) in order to define the natural spaces of \(p\)-integrable functions (Section 4). The norm is substituted in this setting by a \(D^+\)-valued function, and the main properties of the corresponding \(L^p\) spaces are obtained (Hölder type inequality). The paper shows a nice Lebesgue type integration structure based on probabilistic ideas, and the authors claim that there is a great class of applications in physics, as for example the modeling of electric charges distributed on a surface or the mass density in a material when the measurements of these quantities are affected by errors.