OctoInformBeta

OpenAI preprint, September 2026 · §2. Weak compactness and a minimal invariant set

Lemma 2.4 [Compact detector]: One functional separates an accurate approximate fixed point from a prescribed nonempty norm-compact subset of the selfmap’s domain KK

Compact detection near approximate fixed points.

Hover over a blue or green term in the text, or a node in the graph, to find it in the other. Select it to read its declaration both in Lean and in words.

The Setting

Let XX be a real Banach space, and let X*X^* be its continuous real dual with the operator norm. For a selfmap T:K→KT:K\to K of a subset K⊆XK\subseteq X, the fixed-point defect at y∈Ky\in K is ∥Ty−y∥\|Ty-y\|. The issue is the position of points with small defect relative to norm-compact subsets of KK.

A selfmap T:K→KT:K\to K is nonexpansive when it satisfies (1), with both norms taken in the original ambient space: ∥Tx−Ty∥≤∥x−y∥(x,y∈K).\|Tx-Ty\|\leq\|x-y\|\qquad(x,y\in K).(1) A subset s⊆Xs\subseteq X is invariant under TT if TT maps every point of K∩sK\cap s into ss.✦ The pair (K,T)(K,T) satisfies minimal closed convex invariance if every nonempty, norm-closed, convex invariant subset of KK equals KK.✦

Write XwX_{\mathrm w} for the weakly topologized ambient space and ι:X→Xw\iota:X\to X_{\mathrm w} for the canonical identification. The space KwK_{\mathrm w} is the copy ι(K)\iota(K) of KK, equipped with the topology induced from XwX_{\mathrm w}.✦ A subset K⊆XK\subseteq X is weakly compact if ι(K)\iota(K) is compact in XwX_{\mathrm w}.✦

Minimal invariant set data (K,T)(K,T) in XX are a pair consisting of a nonempty, norm-closed, bounded, convex, norm-separable and weakly compact subset K⊆XK\subseteq X, together with a nonexpansive selfmap T:K→KT:K\to K that fixes no point and satisfies minimal closed convex invariance.✦

The closed difference span YY of KK is the closed real linear subspace in (2); BYB_Y is its whole closed unit ball, viewed as a subset of XX: Y=span⁡ℝ(K−K)¯,BY={x∈Y:∥x∥≤1}.Y=\overline{\operatorname{span}_{\mathbb R}(K-K)}, \qquad B_Y=\{x\in Y:\|x\|\leq1\}.(2) The closure and the ball in (2) use the original ambient norm.

Normalized minimal invariant data (K,T)(K,T) in XX are minimal invariant set data with diam⁡K=1\operatorname{diam}K=1, together with a sequence (zj)j∈ℕ(z_j)_{j\in\mathbb N} having norm-dense range in KK, metrizability of KwK_{\mathrm w}, and weak compactness of BYB_Y.✦

The Claim

Theorem 1 (Uniform diametrality, (3)). Let (K,T)(K,T) be normalized minimal invariant data. Let a∈Ka\in K and η>0\eta>0. There is δ>0\delta>0 such that every y∈Ky\in K with ∥Ty−y∥≤δ\|Ty-y\|\leq\delta satisfies 1−η≤∥y−a∥.1-\eta\leq\|y-a\|.(3)✦

Theorem 2 (Compact detector, (4)). Let (K,T)(K,T) be normalized minimal invariant data, let E⊆KE\subseteq K be nonempty and norm-compact, and let ε>0\varepsilon>0. There is δ>0\delta>0 such that, for every y∈Ky\in K with ∥Ty−y∥≤δ\|Ty-y\|\leq\delta, there is f∈X*f\in X^* satisfying ∥f∥≤1,1−ε≤f(y−z)(z∈E).\|f\|\leq1, \qquad 1-\varepsilon\leq f(y-z)\quad(z\in E).(4)✦

What Is Checked

The claims are stated over normalized minimal invariant data in a real Banach space, and compact detection additionally takes a nonempty norm-compact subset E⊆KE\subseteq K.