Henstock–Kurzweil Integral Overview

• The Henstock–Kurzweil integral is a generalized integration method using adaptive gauges and tagged partitions to handle singular and oscillatory functions.
• It unifies and extends classical Riemann and Lebesgue integrals, providing precise versions of the Fundamental Theorem of Calculus even for nonabsolutely convergent functions.
• The framework supports advanced generalizations in multidimensional analysis, Banach space theory, and geometric measure theory, enabling robust analytical tools.

The Henstock–Kurzweil integral, also known as the gauge or generalized Riemann integral, is a nonabsolute integration theory that strictly contains both the Riemann and Lebesgue integrals in one and higher dimensions. Its core idea is to generalize the partitioning process underlying the classical Riemann integral by allowing locally adaptive “gauges,” significantly expanding the scope of integrable functions and enabling a precise and powerful version of the Fundamental Theorem of Calculus. The theory underlies many recent generalizations of integration in analysis, distribution theory, Banach space-valued integration, and geometric measure theory, and provides a unified analytic structure for treating highly oscillatory, unbounded, or singular functions across Euclidean and more abstract spaces.

1. Foundational Definition and Main Construction

The one-dimensional Henstock–Kurzweil integral (HK integral) is defined on a compact interval $[a,b]\subset\mathbb{R}$ by means of tagged partitions and gauges. A gauge is any map $\delta:[a,b]\to(0,\infty)$. A tagged partition is a finite set $\{([x_{i-1},x_i],t_i)\}_{i=1}^N$ with $a=x_0<\cdots<x_N=b$ and $t_i\in[x_{i-1},x_i]$. The partition is $\delta$-fine if $x_i-x_{i-1}<\delta(t_i)$ for all $i$.

A real-valued function $f$ is HK integrable on $[a,b]$ if there exists $I\in\mathbb{R}$ such that for every $\epsilon>0$ there is a gauge $\delta$ for which, for any $\delta$-fine tagged partition,

$$\left| \sum_{i=1}^N f(t_i)(x_i - x_{i-1}) - I \right| < \epsilon.$$

This definition subsumes the classical Riemann approach (which uses uniform gauges) and permits the mesh near troublesome points (e.g., singularities or discontinuities) to be as fine as necessary. In higher dimensions, as in the continuous primitive approach(Talvila, 2019), the fundamental object is an $n$-fold generalization using rectangles or bricks, with the integral grounded in properties of continuous primitives vanishing on the boundary at $-\infty$ in each coordinate.

2. Key Properties and Structural Results

2.1 Linearity, Additivity, and Order

The HK integral is linear: if $f,g$ are integrable and $\alpha,\beta\in\mathbb{R}$, then so is $\alpha f+\beta g$, and $$\int (\alpha f+\beta g) = \alpha\int f + \beta\int g$$. Additivity holds over adjacent intervals, and the integral respects the order structure: $f\leq g$ implies $\int f \leq \int g$(Henstock, 2016).

2.2 Completion and Norm Structures

The primitives of HK-integrable functions admit an isometric isomorphism with the Banach space of continuous functions under the uniform (Alexiewicz) norm. This norm is defined, for an integrable $f$ with primitive $F$ ($F(x)=\int_a^x f$), by $\|f\|_A = \sup_{x\in[a,b]}|F(x)|$ (generalized to $n$-dimensions by $\|f\|_A = \|F\|_\infty$)(Talvila, 2019, Morales et al., 2020). The closure of $L^1$ and of the HK-integrable functions with respect to this norm yields the Banach space $D_{HK}$ of HK-integrable distributions(Morales et al., 2020).

2.3 Relation to Other Integrals

• Riemann and Lebesgue Integrals: Every function that is Riemann- or Lebesgue-integrable is also HK-integrable, and the integrals coincide. The inclusion is strict: there exist HK-integrable functions not integrable in the Lebesgue sense, notably those whose absolute value is non-integrable, and all derivatives on $[a,b]$ are HK-integrable even if unbounded(Henstock et al., 2015, Julia, 2019).
• Dunford-Henstock-Kurzweil Integrals: The DHK integral extends HK-integration to Banach-space valued functions and multidimensional Polydomains using additive interval functions and negligible variation criteria on set boundaries, yielding a true multidimensional extension(Kaliaj, 2018).
• Henstock–Kurzweil–Stieltjes, Hake, and Dunford Integrals: The HK/Stieltjes, Hake–HK, and DHK/DM integrals generalize the gauge integral to functions with respect to arbitrary interval functions, sets with negligible boundaries, distributions, or Banach-space settings(Kaliaj, 2018, Kaliaj, 2018, Morales et al., 2020).

3. Convergence Theorems and Calculus Toolset

3.1 Fundamental Theorem of Calculus

The classical form holds: if $F$ is differentiable on $[a,b]$ with derivative $f$, then $f$ is HK-integrable and $\int_a^b f = F(b) - F(a)$; conversely, the indefinite integral $F(x)=\int_a^x f$ is continuous and differentiable almost everywhere with $F'(x) = f(x)$ almost everywhere(Henstock, 2016, Julia, 2019, Talvila, 2019).

3.2 Limit Theorems

• Monotone Convergence: If $f_n\uparrow f$ and $\sup_n\int f_n<\infty$, then $f$ is integrable and $\int f = \lim\int f_n$.
• Dominated Convergence: Given $|f_n|\leq g$ for an integrable $g$ and $f_n\to f$ pointwise, then $f$ is integrable and $\int f_n\to \int f$(Henstock et al., 2015, Henstock, 2016, Paxton, 2016).
• Uniform Convergence: Uniform limit of integrable functions is integrable, and the limit of the integrals equals the integral of the limit(Paxton, 2016, You et al., 2017).

3.3 Other Analytical Tools

• Integration by Parts: For $f$ integrable and $g$ of bounded variation, the integration by parts formula holds exactly as in Riemann–Stieltjes theory(Talvila, 2019, Morales et al., 2020).
• Change of Variables: Valid for strictly monotone, differentiable substitutions; generalized in higher dimensions to coordinate transformations preserving rectangles(Talvila, 2019).
• Hölder-type and Mean Value Inequalities: Precise estimates extend to multi-parameter settings using Hardy–Krause variation for multidimensional cases(Talvila, 2019).

4. Generalizations and Multidimensional Extensions

4.1 Banach Space and Multidimensional Theory

HK integration admits a robust Banach-space generalization for functions $f:G\to X$ ($X$ Banach), with vector-valued interval sums, local gauges, and descriptive characterizations based on extensions by zero and local integrability(Kaliaj, 2018, Kaliaj, 2018). Both Hake–HK and Dunford–HK integrals enable integration over arbitrary measurable sets with negligible boundaries, including unbounded or highly disconnected domains.

4.2 Distributional and Fractional Calculus

The HK framework extends to the space of distributions $D_{HK}$, with the integral defined for every distribution that is the distributional derivative of a continuous primitive. This enables generalized integration and convolution, a semigroup structure for Riemann–Liouville fractional integrals, solution of generalized Abel equations, and compatibility with fractional calculus and Fourier transforms(Morales et al., 2020).

4.3 Abstract and Geometric Extensions

Recent work generalizes the HK integral to spaces beyond Euclidean, notably to compact metric spaces with arbitrary Borel measures, using fine tagged partitions and nets of simple valuations, yielding the $D_\mu$-integral that recovers the HK and Lebesgue integral on $[0,1]$ and admits non-absolutely integrable singular functions (e.g., on Cantor space)(Edalat, 5 Mar 2025). In geometric measure theory, gauge-integration is adapted to 1-dimensional integral currents, replacing intervals by indecomposable currents and Riemann sums by sums over gauge-fine families of pieces, providing a generalized FTC in this setting(Julia, 2019).

5. Characterizations and Examples

5.1 Characterizations

The HK integrable functions can be characterized in multiple ways:

• Cauchy–Darboux: The set of all gauge-fine Riemann sums is Cauchy.
• Primitives: $f$ is HK-integrable iff it admits a continuous primitive $F$ such that $F(b)-F(a) = \int_a^b f$.
• Sequential and Topological: The HK integral admits fully sequential and topological reformulations, making it adaptable to generalized spaces and net convergence(Paxton, 2016).
• $L^r$-Henstock–Kurzweil Integrals: The $HK_r$ integral, based on $L^r$-derivative recovery on Lusin classes, generalizes the HK integral in the sense that $f$ is $HK_r$ integrable iff there exists a primitive $F \in ACG_r$ with almost everywhere $F' = f$(Musial et al., 2023).

5.2 Illustrative Examples

• Unbounded Derivatives: $f(x)=2x\sin(1/x^2)-(2/x)\cos(1/x^2)$ on $(0,1]$ is HK-integrable but not Riemann or Lebesgue-integrable.
• Oscillatory Singularities: $f(x) = x^{-2}\sin(1/x^2)$ on $(0,1]$ is HK-integrable, $\int_0^1 f = 0$, but $|f|$ is not Lebesgue-integrable.
• Nonabsolutely Integrable Functions: The indicator of rationals (Dirichlet’s function) and Thomae’s function are HK-integrable but not Riemann-integrable.
• Fractal and Abstract Domains: There are functions on the Cantor space that are HK-integrable with respect to suitable measures but not Lebesgue-integrable(Edalat, 5 Mar 2025).

Integral/Class	Characterization	Key Property
Henstock–Kurzweil (HK)	Adaptive gauges, tagged partitions	Integrates strictly more than Riemann/Lebesgue
HK-integrable distributions	Primitives in Banach space $C_0$	Alexiewicz norm, completion of $L^1$
Hake–HK / DHK	Partition-based, negligible boundaries	Integrates on arbitrary $G$ with $|G\setminus G^o|=0$
$D_\mu$-integral	Net of fine tagged partitions	Extends HK to compact metric spaces

6. Analytical and Applied Significance

The Henstock–Kurzweil integral provides a flexible integrative framework that achieves a "Riemann-like" and nonabsolute integration structure, which is critical for several technical applications:

• Resolving the failure of traditional integration theories with derivatives and singularities.
• Establishing robust versions of the FTC without imposing absolute integrability.
• Enabling nonabsolutely convergent integration on manifolds, Banach spaces, and abstract metric domains.
• Supporting operational calculus, convolution, and Fourier analysis for generalized and distributional objects(Morales et al., 2020, Talvila, 2019, Edalat, 5 Mar 2025).

The HK integral's adaptability has spurred diverse generalizations, rigorous Banach norm completions, geometric measure-theoretic constructions, and integration on time scales and nontrivial topological or measure environments, unifying several forms of advanced analysis under a common, powerful framework.