Adjoint classes of Kurzweil-Stieltjes integrable functions (Q1750847)

One of the most interesting topics in Stieltjes integration, together with (of course) integration-by-parts formulae, are adjoint classes of functions. It is of interest for its own, also it often finds application in characterising the dual for some normed spaces of functions. The topic has been initiated, apparently, by \textit{H. Chen} in three consecutive articles [Real Anal. Exch. 23, No. 1, 235--240 (1998; Zbl 0944.26017); ibid. 24, No. 1, 139--147 (1999; Zbl 0945.26013); ibid. 26, No. 1, 421--427 (2001; Zbl 1023.26007)], where a number of pairs of adjoint classes are given in the frame of Riemann, Young, and Lebesgue integration approach. So far, at least up to reviewer's knowledge, no adjoint pair has been provided for the Kurzweil (Kurzweil-Henstock) variant of Stieltjes integration. A somewhat surprising result by \textit{M. Tvrdý} [Mem. Differ. Equ. Math. Phys. 25, 1--104 (2002; Zbl 1081.34504)] suggested two candidates in this connection -- it turns out that \((K)\int_a^bf\,dg\) exists provided one of \(f\), \(g\) is of bounded variation and the other one is regulated (the surprising part is with \(f\) being of bounded variation). The article under review starts here. Tvrdý's result [loc. cit.] is completed by showing (Proposition 4.3) that if a function is (Kurzweil) integrable with respect to all regulated functions, it is necessarily of bounded variation. This results in a pair of adjoint classes for the Kurzweil integral: VB-functions (as integrands) \& regulated functions (as integrators). A simple Example 3.2 shows that, despite symmetricity of Tvrdý's result [loc. cit.], in the reverse order these classes are not adjoint (and even more: no class of integrands is adjoint to VB-functons).