Weak Minimizing Property

• Weak Minimizing Property is a structural minimality principle asserting that relaxed local minimality—measured through weak topologies, local exchanges, or subsequence criteria—can imply global minimality.
• It is applied across diverse fields such as vector optimization, Banach space operator theory, geometric measure theory, discrete convex analysis, and weak K.A.M. dynamics.
• The property underlies dual variational inequality characterizations (Stampacchia and Minty), ensuring norm-attainment, stability under weak convergence, and guiding algorithmic design.

The weak minimizing property (abbreviated in several contexts as WmP) is a family of structural minimality notions that arise across variational analysis, operator theory, optimization in Banach spaces, set-valued and vector optimization, geometric measure theory, and discrete convex analysis. Broadly, the property encapsulates the principle that a certain relaxed (non-strict) form of local minimality, sometimes expressed in terms of weak topologies, local exchanges, or minimization "along subsequences," yields global minimality or rules out strictly superior candidates. The property is central for dual characterizations (Stampacchia and Minty variational inequalities), norm-attainment phenomena, and connections with scalarization and regularization.

1. Weak Minimizing Property in Set Optimization and Vector Optimization

Let $X$, $Y$ be locally convex Hausdorff spaces, and $C \subset Y$ a closed convex cone with non-empty interior. For a set-valued map $F: X \to \mathcal{P}(Y)$ and a weak$^*$-compact base $W^*$ of the positive dual cone, scalarizations $$\varphi_{F, w^*}(x) = \inf\{ w^*(y) : y \in F(x) \}$$ are used.

A point $x_0 \in \operatorname{dom} F$ is a weak minimizer (or weak-scalarized/l-minimizer) of $F$ if either $F(x_0) + C = Y$ or

$$\forall\,x \in X,\, \exists\, w^* \in W^*:\quad \varphi_{F, w^*}(x_0) \leq \varphi_{F, w^*}(x) > -\infty.$$

For compact-valued $F(x)$, this notion coincides with classical weak minimality and, when $F$ is single-valued and vector-valued, with weak Pareto optimality.

This weak minimality is dual to classic efficiency in vector optimization:

• $x_0$ is efficient for $f$ ($F(x) = \{ f(x) \}$) if there is no $x$ with $f(x)$ strictly dominating $f(x_0)$.
• $x_0$ is weakly efficient (weak Pareto optimal) if for all $x$, $f(x_0) \notin f(x) - \operatorname{int} C$.

In convex set-valued contexts, the weak minimizing property admits characterization via variational inequalities for the scalarizations, leading to equivalence with vector efficiency via Stampacchia and Minty VIs (Crespi et al., 2014, Crespi et al., 2014).

2. Duality via Scalarized Variational Inequalities

For set-valued optimization, two principal classes of VI characterize (under regularity/convexity):

(a) Stampacchia-type VI (SVI):

For every $x \in \operatorname{dom} F$ and every $w^* \in W^*$,

$$D^+ \varphi_{F, w^*}\bigl(x_0; x - x_0\bigr) \geq 0,$$

with $D^+$ the lower Dini directional derivative. For convex $F$, SVI are both necessary (under mild assumptions such as radial pseudoconvexity) and sufficient (Crespi et al., 2014).

(b) Minty-type VI (MVI):

$$\forall x \in X,\, \forall w^* \in W^*: \quad D^+ \varphi_{F, w^*}(x; x_0 - x) \leq 0$$

Under further regularity and convexity (radially upper-Hausdorff-continuous, pseudoconvex and pseudoconcave scalarizations), this is sufficient; for convex problems, it also becomes necessary, and the SVI and MVI are equivalent to the (w-Min) property (Crespi et al., 2014, Crespi et al., 2014).

This duality structure is summarized in the implication diagram:

$$\boxed{ \text{Weak minimizer} \Longleftrightarrow \text{Weak Stampacchia VI} \Longleftrightarrow \text{Weak Minty VI} }$$

For set-valued mappings in conlinear spaces $\mathcal{G}(Z, C)$ (closed, convex, $C$-upper sets in $Z$), analogous weak minimizing properties are defined via lattice orderings, scalarizations, and Dini derivatives, with a similar cascade of equivalences as above (Crespi et al., 2014).

3. Weak Minimizing Property in Banach Space Operator Theory

Given Banach spaces $X, Y$ and a bounded linear operator $T \in \mathcal{L}(X, Y)$, the minimum modulus is

$$m(T) = \inf \{ \|T x\| : x \in S_X \}, \quad S_X = \{ x \in X : \|x\| = 1 \}.$$

The operator $T$ is bounded below iff $m(T) > 0$.

Define a minimizing sequence as $(x_n) \subset S_X$ with $\|T x_n\| \to m(T)$. The pair $(X, Y)$ has the weak minimizing property (WmP) if whenever $T$ admits a minimizing sequence not weakly null, then $T$ attains its minimum modulus, i.e., $\exists x \in S_X$, $\|T x\| = m(T)$ (Han, 24 Jan 2026):

$$(X,Y)\text{ has WmP} \iff \forall\,T\in\mathcal L(X,Y),\; [\exists\,(x_n)\not\rightharpoonup 0,\;\|T x_n\|\to m(T)] \Rightarrow\exists\,x\in S_X,\|T x\|=m(T)$$

In particular, pairs such as $(\ell_p, L^p[0,1])$ ($2 \leq p < \infty$) and certain direct sums do satisfy WmP, while $(\ell_1, \ell_p)$, $(c_0, \ell_p)$, etc., do not. WmP is the minimum-modulus analog of the weak maximizing property and connects to norm-attainment, denseness of minimum-attaining operators, and compact-perturbation properties (Han, 24 Jan 2026).

4. Weak Minimizing Property for Quasiminimizing Sequences in Geometric Variational Problems

Within geometric measure-theoretic frameworks, the weak minimizing property asserts that if a sequence of sets $(E_k)$ is "almost minimal" with respect to an admissible energy functional under a class of deformations (e.g., sliding boundary, Plateau-type constraints), and if the associated energy measures $I\llcorner E_k$ converge weakly, then the limiting set is itself a quasiminimal set with similar regularity (Labourie, 2020).

Concretely, for $X \subset \mathbb{R}^n$, $E_k$ closed, $H^d$-locally finite, $\Gamma \subset X$ closed (boundary), and an admissible energy $I$:

• If $E_k$ is $(\kappa, h, s)$-quasiminimizing (with precise "deficit" inequalities under deformations), and $I\llcorner E_k \rightharpoonup \mu$, then the limit set $E = \operatorname{supp} \mu$ is $(\kappa+h, h, s)$-quasiminimal, and the measure satisfies bounds $$I\llcorner E \leq \mu \leq (\kappa + h) I\llcorner E$$.

The weak minimizing property thus justifies the stability of almost-minimizing sets under variational convergence, enabling passage from approximating sequences to limit objects with geometric and energy regularity (Labourie, 2020).

5. Weak Minimizing Property in Discrete Convex Analysis

For functions $f: \mathbb{Z}^n \to \mathbb{R} \cup \{+\infty\}$ that are $M^\natural$-convex or semi-strictly quasi-$M^\natural$-convex (SSQ $M^\natural$), the weak minimizing property states that global minimality is characterized by local minimality with respect to elementary exchange moves:

A point $x^*$ is a minimizer if and only if for all $i, j$,

$f(x^* - \chi_i + \chi_j) \geq f(x^*)$

where $\chi_i$, $\chi_j$ are unit vectors or zero.

Despite the weakened exchange axiom in the SSQ $M^\natural$ case, the same local-to-global implication holds. This property enables verification of global minimizers via simple local optimality checks and supports descent-based algorithms (Murota et al., 2023).

6. Weak Minimizing Property in Weak K.A.M. Theory and Hamilton-Jacobi Dynamics

In the context of weak K.A.M. theory (continuous and discrete), the weak minimizing property manifests as the calibration of subsolutions by minimizing orbits or random walks that realize the minimum action. For twist maps and Tonelli Hamiltonians, weak K.A.M. solutions $u_c$ are such that their backward calibrated semi-orbits precisely equate the change in $u_c$ to the summed Lagrangian action plus the corrector term.

This property supports the pseudograph foliation of phase-space and underlies the Lipschitz dependence of weak K.A.M. solutions on cohomological parameters, vertical ordering of pseudographs, and the identification of Mather and Aubry sets. Discrete analogs in grid-based or random-walk Hamilton-Jacobi discretizations exhibit calibration identities that converge to their continuous counterparts under suitable scaling (Arnaud et al., 2022, Soga, 2020).

7. Summary Table: Weak Minimizing Property Across Contexts

Mathematical Context	Formalization of Weak Minimizing Property	Representative Paper
Set/Vector Optimization	Scalarized VI: $$\exists w^*:\ \varphi_{F,w^*}(x_0) \leq \varphi_{F,w^*}(x)$$	(Crespi et al., 2014, Crespi et al., 2014)
Operator Theory (Banach)	Non-weakly null minimizing sequences force attains minimum modulus	(Han, 24 Jan 2026)
Geometric Measure Theory	Weak limit of quasiminimizing sequence is quasiminimal	(Labourie, 2020)
Discrete Convex Analysis	Local exchange optimality implies global minimality	(Murota et al., 2023)
Weak K.A.M./Hamilton-Jacobi	Calibrating semi-orbits/paths realize minimal actions	(Arnaud et al., 2022, Soga, 2020)

Each instance features a distinct yet structurally similar local-to-global minimality principle, often dual to maximizing properties and intimately linked to notions of calibration, convexity, and variational characterizations.

---

The formal landscape provided by the weak minimizing property enables a unified treatment of minimality, variational principles, and efficiency, spanning infinite-dimensional spaces, discrete lattices, operator spaces, and geometric settings. Its centrality is exemplified by equivalence theorems for scalarized variational inequalities, norm-attainment phenomena, stability under weak convergence, and algorithmic implications for local-global optimality checks. Continued research explores geometric extensions, characterization in broader Banach pairs, discrete and nonlinear generalizations, and deeper connections with duality and calibration principles (Crespi et al., 2014, Crespi et al., 2014, Han, 24 Jan 2026, Labourie, 2020, Murota et al., 2023, Arnaud et al., 2022, Soga, 2020).