Primitives of Uniformly Continuous Functions

• Uniformly continuous functions are defined by a uniform δ for all ε > 0, and their primitives explore how antiderivatives retain analytical properties.
• Sufficient conditions like piecewise convexity are shown to guarantee absolute continuity, bridging uniform continuity with stronger regularity classes.
• The study also examines algebraic, topological, and computational structures, including extension theorems and effective algorithms for function approximation.

A uniformly continuous function is a function between metric or uniform spaces that admits a uniform modulus of continuity: for every $\varepsilon > 0$ there is a $\delta > 0$ such that, for all pairs $x, y$ with $d(x, y) < \delta$, $|f(x) - f(y)| < \varepsilon$. The study of the primitives of uniformly continuous functions addresses how foundational operations—such as forming antiderivatives, extension, composition, and algebraic closure—interact with uniform continuity, and under what structural, topological, or algebraic constraints stronger analytic regularity or algorithmic properties can be obtained. This involves detailed characterizations, sufficient conditions, and limitations, as well as applications to integration, differentiation, topological structure, and computational complexity.

1. Sufficient Conditions for Absolute Continuity of Uniformly Continuous Functions

Absolute continuity is a strictly stronger regularity property than uniform continuity; it ensures almost everywhere differentiability and allows for the reconstruction of a function from its derivative via the Fundamental Theorem of Calculus. Normally, uniform continuity does not guarantee absolute continuity: classical counterexamples include the Cantor function, which is uniformly but not absolutely continuous.

A notable sufficient condition is given by the following theorem: if $f$ is a uniformly continuous and piecewise convex function on an interval $I_{a,b} \subset \mathbb{R}$, then $f$ is absolutely continuous on $I_{a,b}$ (Yang et al., 2010). Here, $f$ is called piecewise convex if there exists a finite partition $P = \{a_0, a_1, \ldots, a_N\}$ such that on each $[a_i, a_{i+1}]$, $f$ is either convex or concave.

Key steps in the proof exploit monotonicity of the difference $G_\sigma(x) = |f(x + \sigma) - f(x)|$ on segments of monotonicity and convexity, using it to control total variation and hence to establish absolute continuity from uniform continuity under finite convexity changes. This bridges the classical implication "absolute continuity $\Rightarrow$ uniform continuity" by identifying a significant class where the converse holds.

Examples in the paper clarify these boundaries: $f(x) = \sqrt{x}$ on $[0, c]$ satisfies the condition (uniformly continuous and convex), so it is absolutely continuous despite not being globally Lipschitz; in contrast, $f(x) = x^2 \sin(1/x)$ on $[0, 1]$ is absolutely continuous but not piecewise convex, demonstrating strict inclusion.

2. Algebraic and Topological Structure: Rings, Lattices, and Approximations

Uniformly continuous functions do not always form a ring under pointwise multiplication in arbitrary metric spaces. However, for many natural spaces (including compact and certain locally compact spaces), their set $U(X)$ of all real-valued uniformly continuous functions admits additional algebraic and order structure.

For any metric space $X$:

• $U(X)$ is a ring if and only if every subset $A\subset X$ is either Bourbaki-bounded (every $f\in U(X)$ is bounded on $A$) or contains an infinite uniformly isolated subset (some $\delta >0$ with $d(a,x)\geq \delta$ for $a$ in an infinite $F\subset A$ and every $x\in X$) (Sánchez, 2017).
• Equivalent conditions are established regarding the structure of "tails" of functions—sets where $|f(x)| > k$—requiring them to be uniformly isolated for all $f\in U(X)$ (Bouziad et al., 2019).
• On normed vector spaces of dimension one, there is an isomorphism between the lattice of Lipschitz and bounded Lipschitz functions; in higher dimensions, this property fails (Sánchez et al., 2019). The lattice $U(X)$ can be equipped with a unital $f$-ring structure via a modified product $M(f, g)(x) = f(x)\,g(x)/A(x)$, provided there exists a dominating function $A$ as described.

These results are crucial for preserving uniform continuity under algebraic operations and allow analytic constructions (integration, antiderivatives) to remain within the same functional class, provided the underlying space satisfies strict separation or boundedness properties.

3. Extension and Approximation Theorems

Classical extension theorems (Lebesgue, Tietze, Dugundji) ensure that continuous functions defined on closed subsets can be extended continuously to the whole space. Many explicit formulas, such as Hausdorff’s and various Tietze-like constructions based on the distance function, not only preserve continuity but also uniform continuity (Gutev, 2020). This preservation enables construction of extension operators that serve as primitives for spaces of uniformly continuous functions, essential for functional approximation, partition of unity constructions, and embedding theorems.

The simultaneous extension property—where one operator extends all bounded or special bounded functions in a norm-preserving, order-isotone (monotone) fashion—provides a robust analytical toolkit. In such frameworks, nonlinear extension operators arise naturally; while no linear operator preserves uniform continuity, sublinear and isotone extensions do.

In the constructive mathematics setting, the strong continuity principle (SC) holds if and only if a variant of the monotone $\Pi^0_1$ fan theorem holds (Kawai, 2018). SC asserts that every pointwise continuous function from a complete separable metric space to a metric space is uniformly continuous near each compact image. This logical equivalence connects analytic extension (and "niceness" of function primitives) with logical and combinatorial metamathematics.

On the measure-theoretic side, every Borel function on a Polish space equipped with finite Borel measure can be approximated in measure (and in $L^p$ for bounded functions) by sequences of bounded, uniformly continuous functions (Chou, 2020). This generalizes Lusin's theorem and underpins convergence and integration results for non-locally compact spaces.

4. Topological and Descriptive Structure of Function Spaces

The internal topological structure of spaces of uniformly continuous functions has been analyzed using tools from infinite-dimensional topology. For a separable, locally compact, regular Borel metric measure space $(X, d, M)$, the subset of $L^p(X)$ consisting of (almost everywhere) uniformly continuous functions, $C_u(X)$, is homeomorphic to $c_0$, the space of sequences converging to $0$ in the pseudo interior of the Hilbert cube (Koshino, 2020). This results from the absorption property and universality arguments using Z-set theory and strong $M_2$-universality.

In lattice-theoretic terms, the space of homomorphisms from $U$ (the lattice of uniformly continuous functions, e.g. on the half-line $[1,\infty)$) to $\mathbb{R}$ has a fine structure: all nonzero homomorphisms are either evaluations at points (local primitives) or outer functionals sensitive to asymptotic behavior, often constructed with ultrafilters (Sánchez, 2019). The resulting topological space of homomorphisms is a compactification with an explicit quotient description, bridging functional analysis and descriptive set theory.

5. Computational and Algorithmic Complexity

The question of computability and computational complexity of uniformly continuous functions and their primitives is addressed via rational presentations of complete metric spaces (Lombardi et al., 19 Feb 2025). In particular:

• Uniformly continuous real functions on $[0,1]$ with the supremum norm are amenable to distinct "presentations": Ko–Friedman codes (based on Turing machines), Boolean/arithmetic circuits, and polynomial (Weierstrass) expansions.
• These presentations are polynomial-time equivalent for evaluation, but computing the supremum norm is NP-hard in several codes, even co-NP-complete for membership testing.
• The Weierstrass (polynomial) presentation allows linear-time verification of membership, making it more practical for tasks such as checking uniform approximation or computing primitives.
• The complexity of forming primitives (antiderivatives) efficiently depends on the chosen representation; presentations facilitating fast norm computation are preferable for numerical analysis of integration.

Extending this, there exist effective algorithms (with explicit error bounds) for uniform approximation by polynomials or rational functions (e.g., Newman's theorem for approximating $|x|$) with implications for primitive computation in polynomial time.

6. Primitives of Continuous and Uniformly Continuous Functions: Polynomial Construction

An elementary construction of primitives for continuous functions on $[0,1]$ uses the sequence of Bernstein polynomial approximations $f_n$ and their explicit polynomial primitives $F_n$ (Lundström, 2022). By uniform convergence of $f_n$ to $f$ and control over the derivatives $F_n'$ (since $F_n' = f_n$), it is shown that the uniform limit $F$ of the $F_n$ is an antiderivative of $f$, i.e., $F'(x) = f(x)$ almost everywhere. This method fully leverages uniform continuity to guarantee uniform convergence, thus providing a direct pathway from uniform continuity to the effective constructibility of primitives.

7. Connections, Limitations, and Open Problems

Uniform continuity is strictly weaker than absolute continuity or differentiability, as demonstrated by the existence of functions such as the Cantor function. Sufficient conditions, such as piecewise convexity plus uniform continuity, enable passage upwards in regularity.

Topological restrictions and set-theoretic assumptions come into play, as highlighted by recent work on monotone functions (Pol et al., 28 Feb 2025): for every non-decreasing $f: [0,1] \to [0,1]$, under certain hypotheses on cardinal characteristics (e.g., $\mathfrak{d}^* < \mathfrak{c}$, $\mathfrak{c}$ regular), every subset of cardinality $\mathfrak{c}$ contains a subset of the same cardinality where $f$ is uniformly continuous. This explores the subtle dependence of uniform continuity on global set-theoretic and topological invariants.

Further open problems include:

• Characterizing when products or unions of spaces preserve "good" algebraic or analytic properties of uniform continuity.
• Determining the exact boundary between uniform and absolute continuity under additional regularity hypotheses.
• Developing refined extension and approximation operators that optimize for algorithmic (complexity) and analytic (modulus of continuity) properties.
• Extending continuous primitive constructions to higher-order forms, manifolds, and non-Euclidean settings, as in recent advances on $L^n$-controlled continuous primitives for differential forms (Baldi et al., 2024).

Taken together, the primitives of uniformly continuous functions encompass analytic, algebraic, topological, and computational aspects, each with sharp results, concrete counterexamples, and ongoing lines of research.