# Thomson's variational measure and nonabsolutely convergent integrals

### Real Analysis Exchange 26 (2000/2001), no. 1, 35-50.

#### Mathematical reviews subject classification: 26A45, 26A39, 26A46, 26A24.

## Abstract

In 1987 Jarnik and Kurzweil (see J.Jarnik and J. Kurzweil, *A general form of the product integral and linear ordinary differential **equations*, Czech. Math. J. **37** (1987), no. 112, 642-659) proved the following result:

A function F:[a,b] → R is AC*G on [a,b] if and only if μ_{F*} (Thomson's variational measure) is absolutely continuous on [a,b] and F is derivable a.e. on [a,b].

But condition "F is derivable a.e. on [a,b]'' is superfluous, as it was shown in Article 17.

In this paper we shall improve this result, from where we obtain an answer to a question of Faure (see C. A. Faure, *A descriptive definition of the KH-Stieltjes **integral*, Real Analysis Exchange **23** (1997-1998), no. 1, 113-124).

Then using Faure's definition for a Kurzweil-Henstock-Stieltjes integral with respect to a function ω, we give corresponding definitions for:

- a Denjoy*-Stieltjes integral with respect to ω,
- a Ward-Perron-Stieltjes integral with respect to ω,
- a Henstock-Stieltjes variational integral with respect to ω,

and we show that the four integrals are equivalent.

Finally, using the notion of AC* with respect to ω (short AC*_{ω} ) and a Lusin condition with respect to ω (short N_{ω} ), we give a Banach-Zarecki type theorem (see S. Saks, *Theory of the integral,* 2nd. rev. ed., vol. PWN, Monografie Matematyczne, Warsaw, 1937).