Extended Metric-Topological Spaces

• Extended metric-topological spaces are defined as triples (X, τ, d) that separate the metric and topological structures, enabling metrics to take infinite values.
• They enhance analysis in infinite-dimensional settings by ensuring compatibility between Lipschitz functions and the underlying topology, crucial for optimal transport and non-smooth analysis.
• Their categorical bicompleteness and generalized uniform structures support universal constructions and diverse applications, from abstract Wiener spaces to TVS-cone metrics.

Extended metric-topological spaces generalize classical metric spaces by allowing metrics to take infinite values and by separating the topology from the metric structure, subject to rigorous compatibility conditions. This framework is essential for the analysis of infinite-dimensional metric-measure spaces, optimal transport, and non-smooth geometric analysis, enabling fine control over analytic and topological properties, even in settings where standard metric spaces are inadequate.

1. Foundational Definitions and Characterization

An extended metric-topological space is defined as a triple $(X, \tau, d)$ with:

• $d: X \times X \to [0, \infty]$ a symmetric extended metric (i.e., $d(x, x) = 0$, $d(x, y) = d(y, x)$, $d(x, z) \leq d(x, y) + d(y, z)$, and possibly $d(x, y) = \infty$).
• $\tau$ a Hausdorff topology on $X$, compatible with $d$ via the initial topology induced by all bounded, $\tau$-continuous, $d$-Lipschitz functions, denoted $\mathrm{Lip}_b(X, \tau, d)$. Precisely, $\tau$ is the coarsest topology making every such $f$ continuous.
• The duality condition: the metric is recovered by

$$d(x, y) = \sup \left\{|f(x)-f(y)| : f \in \mathrm{Lip}_b(X, \tau, d), \mathrm{Lip}(f) \leq 1\right\},$$

so that $d$ and $\tau$ determine each other (Savaré, 2019, Pasqualetto et al., 12 Jan 2026, Pasqualetto et al., 4 Mar 2025).

Key examples include:

• Metric measure spaces $(X, d, m)$ with $d$-topology.
• Banach spaces equipped with the norm metric and the weak or weak* topology.
• Abstract Wiener spaces: $d(x, y)$ is the Cameron–Martin norm if $x-y \in H$ (the Cameron–Martin space), otherwise $+\infty$, with $X$ carrying either strong or weak topology (Savaré, 2019, Pasqualetto et al., 12 Jan 2026).
• Discrete spaces with $d(x,y) = \infty$ for $x \neq y$ and arbitrary $\tau$.

2. Categorical Framework and Universal Constructions

The category $\mathbf{ExtMetTop}$ of extended metric-topological spaces consists of objects $(X, \tau, d)$ and morphisms ("continuous-short maps") that are continuous for $\tau$ and nonexpansive for $d$:

$$d_Y(\varphi(x), \varphi(x')) \leq d_X(x, x'), \quad \forall x, x'\in X.$$

This category is bicomplete: it admits all small limits (products, equalizers) and colimits (coproducts, coequalizers), making it highly stable under universal constructions (Pasqualetto et al., 12 Jan 2026). Bicompact and metrically-complete subcategories are reflective, with functorial compactification and completion procedures.

Notable special cases:

• The subcategory of compact extended metric-topological spaces (compact $\tau$), with a canonical compactification functor analogous to the Stone–Čech compactification (Pasqualetto et al., 12 Jan 2026).
• Metrically complete subcategory via $d$-completion, carried out within compactification.

3. Topological and Uniform Structures

Extended metric-topologies are completely regular, Hausdorff by construction, and the balls $\{y: d(x, y) < r\}$ form a base. The metric may be infinite, splitting $X$ into disconnected components separated by infinite distance. For countable metric families or directed subalgebras, the topology is first-countable and often metrizable by a single extended metric (Savaré, 2019, Taho, 22 Apr 2025).

For spaces with a family of metric-like functions $\{d_i\}_{i \in I}$, a general axiomatic framework produces a topology, (quasi-)uniformity, and—in countable/symmetric cases—a single extended metric:

$$D(x, y) = \sum_{n=1}^{\infty} 2^{-n} \min\{d_{i_n}(x, y), 1\}$$

The induced topology is the coarsest making all $d_i(x,\cdot)$ continuous (Taho, 22 Apr 2025).

4. Analytic Structures: Derivations and Sobolev Spaces

Extended metric-topological measure spaces $\mathbb{X} = (X, \tau, d, \mathfrak{m})$ underpin metric analysis in infinite dimensions (Pasqualetto et al., 4 Mar 2025). The Banach algebra $\mathrm{Lip}_b(X, \tau, d)$ of bounded, $\tau$-continuous, $d$-Lipschitz functions is dense in $L^p(\mathfrak{m})$ for all $p$.

First-order calculus employs Lipschitz derivations: linear maps $$b: \mathrm{Lip}_b(X, \tau, d) \to L^0(\mathfrak{m})$$ obeying the Leibniz rule and classified (e.g., Weaver and Di Marino derivations) by continuity and localization properties. The divergence operator is defined via integration by parts.

Sobolev spaces are constructed via:

• Relaxation (Cheeger energy):

$$\mathcal{E}_p(f) = \inf\left\{\liminf_{n \to \infty} \int (\mathrm{lip}_a(f_n))^p \, d\mathfrak{m} : f_n \to f \text{ in }L^p, f_n \in \mathrm{Lip}_b\right\},$$

leading to $H^{1,p}(\mathbb{X})$.

• Integration-by-parts and derivation duality:

$W^{1,p}(\mathbb{X})$ defined via duality with divergence derivations, with an isometric identification of the predual space:

$$W^{1,p}(X) \cong \left((L^q \times_q \mathrm{Der}^q)/B_q \right)'.$$

• On complete extended metric spaces, $H^{1,p} = W^{1,p} = N^{1,p}$ (Newtonian space) with equality of relaxed slopes.

This analytic machinery extends all known results in classical metric measure spaces and enables differential calculus and gradient flow analysis in highly non-Euclidean contexts (Pasqualetto et al., 4 Mar 2025, Savaré, 2019).

5. Extensions, Generalizations, and Comparisons

Extended metric-topological spaces encompass several further generalizations:

• TVS-cone metric spaces: Distances valued in ordered cones of a locally convex topological vector space, yielding metrizable, paracompact topologies and supporting scalarization techniques for fixed-point theorems and topological comparisons (Abdeljawad et al., 2011).
• Power-set metrics ($P_Z(S)$-metrics): Set-valued distances with the triangle axiom replaced by subset inclusion, reconciling classical and group-valued metrics and generating topologies via “balls” defined by set containment (Jafarpour-Golzari, 2015).
• $\mathcal{F}$-metrics: Axiomatizing distances via non-decreasing functions $f : (0, \infty) \to \mathbb{R}$, producing topologies that are always Hausdorff and first-countable, and supporting generalized fixed-point theory. Metrizability is unresolved and depends on additional properties (Bera et al., 2018).

Spaces equipped with families of metric-like functions, such as those arising in mapping spaces (compact-open topology), spaces of submanifolds, tiling spaces, and embedded graphs, are unified by the extended metric-topological framework, allowing topological, uniform, and diffeological structures to be constructed in parallel (Taho, 22 Apr 2025).

6. Applications and Limitations

Applications to infinite-dimensional geometry include abstract Wiener spaces, Hellinger–Kantorovich spaces, and optimal transport. The bicompleteness of $\mathbf{ExtMetTop}$ guarantees stability under direct or inverse limits, quotients, and extensions (Pasqualetto et al., 12 Jan 2026). The flexibility of separating topology from metric structure is key for handling weak topologies, infinite separation, and non-smooth analysis.

Limitations include:

• Nontrivial metrizability results (e.g., for $\mathcal{F}$-metrics).
• Identifying precise Hausdorff, normal, and compactness conditions in generalized spaces.
• The necessity of reflective and coreflective functors for compactification and completeness (Pasqualetto et al., 12 Jan 2026).

Ongoing research addresses linear isometric extension operators for metrics, the structure of metric spaces of metrics under supremum metrics, and the general classification of uniformity properties in these general contexts (Ishiki, 2024, Savaré, 2019).

7. Comparative Table of Extended Metric Structures

Structure	Metric Type	Topology Derived via
Extended metric-topological	$d: X \times X \to [0, \infty]$	Initial topology from $d$-Lipschitz functions
TVS-cone metric space	$d: X \times X \to E$ (cone in TVS)	Cone-balls, scalarization, seminorm family
$P_Z(S)$-metric space	$d: X \times X \to P(S)$	Set-containment balls in $P_Z(S)$
$\mathcal{F}$-metric space	$d: X \times X \to [0, \infty)$, $f$-modulation	Open balls via $d(x, y) < r$
Extended metric family	$\{d_i\}_{i \in I}$	Basis from $d_i(x, y) < \varepsilon$

Each of these inherently extends the classical metric topology and enables analysis in spaces where traditional metric spaces are insufficient for describing geometric and analytic phenomena.

---

For a rigorous treatment of extended metric-topological spaces and their analytic, categorical, and topological properties, see (Savaré, 2019, Pasqualetto et al., 4 Mar 2025, Pasqualetto et al., 12 Jan 2026, Taho, 22 Apr 2025, Jafarpour-Golzari, 2015), and (Abdeljawad et al., 2011) for cone structures. The categorical bicompleteness and duality properties, as well as the interplay between measure, topology, and metric, are of central importance in current research.