Generalized Compactness in Mathematics

• Generalized Compactness is a framework that extends classical compactness by integrating approximation schemes, measure theoretic methods, and operator techniques across diverse mathematical settings.
• It provides explicit criteria—such as Q-Kolmogorov numbers and μ-compactness—to unify analyses in Banach spaces, convex sets, PDEs, and function space theory.
• Applications span Banach space theory, convex geometry, PDEs, and mathematical physics, offering sharper invariants and new stability phenomena.

A generalized compactness property extends, unifies, or sharpens the classical compactness notion in topological, analytical, or geometric settings. These frameworks capture forms of precompactness, continuity, or sequential compactness under structure-specific constraints, often accommodating non-standard coverings, approximation processes, measure-theoretic properties, or topological operations. Generalized compactness serves as a methodological bridge linking classical results to new phenomena in Banach space theory, convex geometry, function space analysis, PDEs, topology, and mathematical physics.

1. Generalized Compactness via Approximation Schemes

In Banach spaces, a primary formalism for generalized compactness is provided by the concept of a generalized approximation scheme $Q=\{Q_n(X)\}_{n=0}^\infty$ (Aksoy et al., 2013). For $X$ a Banach space, $(Q_n)$ is a sequence of subsets of $X$ satisfying:

• GA1: $Q_0(X)=\{0\}$ and $Q_n(X)$ increases to $X$
• GA2: Scalar multiplication preserves each $Q_n$
• GA3: Minkowski sum $A+B \subset Q_{n+m}(X)$ for $A \in Q_n(X)$, $B \in Q_m(X)$

For a bounded subset $D\subset X$, the $n$-th Q-Kolmogorov number is defined as

$$d_n(D;Q) = \inf\{ r>0 \,|\, D \subset r B_X + A,\, A \in Q_n(X)\}$$

with $B_X$ the unit ball. $D$ is Q-compact if $\lim_{n\to\infty} d_n(D;Q)=0$.

Key implications and examples:

• If $Q_n$ are finite-dimensional subspaces, Q-compactness coincides with relative compactness.
• Nonlinear or infinite-dimensional $Q_n$ (e.g., spaces of $n$-nuclear operators) enable analysis well beyond classical settings.
• Q-compact sets/operators unify the treatment of Kolmogorov widths, entropy numbers, $s$-numbers, and various operator ideals.
• Not every Q-compact operator is compact: in $L^p([0,1]),\, p>2$, the Rademacher projection is Q-compact for an appropriate scheme but not classically compact.

Q-compactness admits alternative characterizations via diagonal sums of order-$c_0$ sequences and by the existence of representing Q-compact maps. Compactness measures such as

$\beta(D;Q) = \lim_{n\to\infty} d_n(D;Q)$

quantify the "distance" to Q-compactness.

2. Generalized Compactness in Convex Sets and Operator Theory

The μ-compactness paradigm, introduced for convex subsets $A$ of locally convex spaces, is defined via probability measure preimages under the barycenter map (Protasov et al., 2010):

$$b: \mathcal{M}(A) \to \mathrm{co}A, \quad b(\mu) = \int_A x\,d\mu$$

$A$ is μ-compact if $b^{-1}(K)\subset \mathcal{M}(A)$ is compact for every compact $K \subset \mathrm{co}A$ (weak topology). This strictly generalizes classical compactness and is preserved under countable products and closed convex hulls.

Significant features:

• μ-compact convex sets generalize key results (e.g., Vesterström–O'Brien theorem: openness of mixture and barycenter maps, continuity of the convex hull operator) formerly valid only in the compact scenario.
• Pointwise μ-compactness (fibers compact) is strictly weaker and insufficient for these properties.
• Noncompact but μ-compact sets include the positive cone in $\ell^1$ and state spaces of quantum statistical models, leading to direct applications in quantum information (e.g., convex-roof extensions, LOCC monotonicity, regularity of entanglement measures).

3. Generalized Compactness for Function Spaces and PDEs

In $L^2$-variational settings, such as the generalized Aviles–Giga functional, compactness of sequences of bounded energy is established through entropy-entropy flux techniques and compensated compactness (Lamy et al., 2022). For the energy

$$I_\varepsilon(m; \Omega) = \int_\Omega \left( \varepsilon |\nabla m|^2 + \frac{1}{\varepsilon}(1-\|m\|^2)^2 \right) dx$$

with divergence-free $m$ and strictly convex norm $\|\cdot\|$, one proves that bounded-energy sequences are $L^2$-precompact, with limits satisfying generalizations of the Eikonal system.

Methodological advances include:

• Uniform entropy production bounds and $H^{-1}$-compactness for divergence-free entropic vector fields.
• Rigidity reduction via Young measures and the exclusion of nontrivial oscillations under strict convexity.
• Extension of L^2 compactness results from the Euclidean to the anisotropic norm case.

Similar themes appear in function spaces of bounded deformation ($GBD$, $GSBD^p$), where generalized compactness results yield a.e. convergence of sequences modulo subtraction of piecewise infinitesimal rigid motions, subject to a Caccioppoli partition (Chambolle et al., 2022).

4. Topological and Abstract Notions of Generalized Compactness

a) Generalized Topologies

In the generalized topology formalism of Delfs–Knebusch (Piȩkosz et al., 2014), a generalized topology is a collection of admissible open families $\mathrm{Cov}_X$ and the corresponding "weak" topology $T(\mathrm{Op}_X)$.

Three nested compactness types arise:

• Topological compactness: every weakly open cover has a finite subcover.
• Absolute compactness: every open cover has a finite subcover.
• Admissible compactness: every admissible cover has a finite subcover.

For weakly Hausdorff spaces, these types coincide under appropriate closure properties. Wallman-type compactifications and their compactness properties are intertwined with the Ultrafilter Theorem, with ZF+UFT (but not ZF) sufficing to guarantee the general theory.

b) Ordinal and Covering Compactness

A further parameterization using covering properties by order-type, not cardinality, is developed as $[\alpha,\beta]$-compactness (Lipparini, 2010). $(X,\mathcal{T})$ is $[\alpha,\beta]$-compact if every cover of length $\beta$ admits a subcover of order-type $<\alpha$.

This yields a stratified compactness hierarchy:

• For cardinals, it recovers two-cardinal compactness.
• For ordinals, it distinguishes topological spaces even when classical cardinal compactness is insensitive.
• Notable transfer and monotonicity properties relate $[\alpha,\beta]$-compactness to that for other parameter pairs.
• For small cardinals ($|X|=\omega$), the notion stabilizes at $\omega^2$.

5. Metric and Topological Variants

Beyond functional and convexity-based generalizations, metric spaces admit further variants:

• Property I: a subset $A\subset X$ has property I if for every open set $U\supset A$, $\inf_{a\in A}d(a, U^c)>0$ (Gupta et al., 2023).
  • Every compact is property I, but property I spaces may not be totally bounded or possess limit points.
  • Property I is stable under uniform homeomorphisms and characterizes when Vietoris and Hausdorff hypertopologies coincide.
• GCC and CCC spaces: "Generalized Compact and Connected" (GCC, no infinite disjoint open covers), and "Compact–Connected Continuous images" (CCC, continuous image of $K\times C$ for compact $K$, connected $C$) extend compactness and connectedness. On $\mathbb{R}$, GCC and CCC coincide; in higher dimensions they diverge (Elez et al., 2024).

6. Generalized Compactness for Sets of Finite Perimeter and Geometric Analysis

For Riemannian manifolds with bounded geometry, generalized compactness captures the limiting behavior of finite-perimeter sets in non-compact settings by compactifying the ambient space with limit manifolds at infinity (Flores et al., 2015):

• A sequence of sets with uniformly bounded volume and perimeter admits a convergent subsequence in the enlarged space $\tilde M = M \bigsqcup_i M_{\infty,i}$.
• This result extends the compactness theory for integral currents and underpins existence and continuity results for isoperimetric profiles in noncompact geometry.

7. Generalized Compactness in Analysis, Probability, and Approach Spaces

Approach theory provides quantitative compactness indices (relative sequential compactness, Lindelöf index, etc.) for abstract "approach spaces," encompassing metric, topological, and probabilistic convergence (Berckmoes, 2015):

• For locally countably generated approach spaces,

$$\mathrm{X}_{\mathrm{rsc}}(A) \leq \mathrm{X}_{\mathrm{rc}}(A) \leq \mathrm{X}_{\mathrm{rsc}}(A) + \mathrm{X}_L(X)$$

and compactness can be characterized in terms of vanishing of these indices.

• In spaces of probability distributions (continuity approach structures), this yields "quantitative Prokhorov theorems" measuring tightness and compactness via explicit escape indices, blending topology with measure-theoretic compactness.

---

Generalized compactness concepts enable the synthesis and extension of classical principles across functional analysis, convexity, metric geometry, PDE, and mathematical physics, while often providing explicit compactness criteria, new stability phenomena, and sharper invariants for structure-dependent convergence and approximation (Aksoy et al., 2013, Protasov et al., 2010, Lamy et al., 2022, Piȩkosz et al., 2014, Lipparini, 2010, Chambolle et al., 2022, Flores et al., 2015, Gupta et al., 2023, Elez et al., 2024, Berckmoes, 2015).