A quasi-additivity type condition and the integral over a BV variety (Q1204314)

Let \(p=(p_ 1,\dots,p_ m)\in E_ m\), \(q=(q_ 1,\dots,q_ k)\in E_ k\), \(K\subset E_ m\), and \(F: K\times E_ k\to E_ 1\) be a function such that (1) \(F\) is bounded and uniformly continuous in \(K\times S\), where \(S\) is the unit sphere in \(E_ k\), (2) \(F(p,tq)=tF(p,q)\) for all \(t\geq 0\), \(p\in K\), \(q\in E_ k\), and \(T: A\to K\) is a continuous variety, where \(A\) is a subset of a topological space. L. Cesari proved that, if \(\{I\}\) is a collection of subsets of \(A\), \(\varphi: \{I\}\to E_ k\) is a set function ``quasi-additive'' and of ``bounded variation'' and \(\omega: \{I\}\to A\) is a ``choice function'', i.e. \(\omega(I)\in I\), then the set function \(\Phi(I)=F(T(\omega(I)),\varphi(I))\) is ``quasi- additive'' and of ``bounded variation'': thus the integral of the calculus of variations \({\mathcal I}=\int_ T F(p,q)\) can be defined as the Burkill-Cesari integral \(BC\int\Phi(I)\). In this paper the authors extend the definition of the integral \(\mathcal F\) to the case where ``\(T\)'' is a variety of ``bounded variation'', not necessarily continuous, and \(F: M\times E\to B\) is a nonlinear parametric integrand (\(M\) is a metric space, \(E\) is a uniformly convex Banach space and \(B\) is a real Banach space) that satisfies the conditions (1) and (2) with \(K\) replaced by \(M\) and \(E_ k\) replaced by \(E\). To this end the authors introduce the ``\(\Gamma\)-quasi-additivity'' condition on the couple of set functions \((p,\varphi)\) and they prove that, if \(\rho: \{I\}\to M\) is an arbitrary set function, \(\varphi: \{I\}\to E\) is a set function of ``bounded variation'' and \((p,\varphi)\) is ``\(\Gamma\)-quasi- additive'', then the set function \(\Phi(I)=F(p(I)\), \(\varphi(I)\) is ``quasi-additive'' and of ``bounded variation'': thus the integral \(\mathcal I\) can still be defined as a Burkill-Cesari integral \(BC\int \Phi(I)\). Therefore the authors show that, even in the new general situation, the integral \(BC\int \Phi(I)\) has a Lebesgue-Stieltjes integral representation in terms of a finite measure \(\mu\), a vector measure \(\nu\) and the Radon-Nikodým derivative \(d\nu/d\mu\). Finally, they present some applications for this new type of integral over a curve of ``bounded variation'', also in the presence of a weight of ``bounded variation''.