OctoInformBeta

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

§2 (setting): weak topology, invariant sets, and the difference ball: Minimality under a selfmap is imposed among nonempty closed convex invariant subsets

Weak topology and the difference ball.

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 normed space and K⊆XK\subseteq X. Let T:K→KT\colon K\to K be a selfmap. Weak compactness concerns the weak topology on KK, whereas the closed linear span of its differences and the unit ball of that span use the ambient norm.

Write X*X^* for the continuous real dual of XX, and XwX_{\mathrm w} for XX equipped with the weak topology σ(X,X*)\sigma(X,X^*). The canonical identification ι:X→Xw\iota\colon X\to X_{\mathrm w} preserves the underlying points. The notation K−KK-K means {x−y:x,y∈K}\{x-y:x,y\in K\}.

The Definition

Definition 1. The copy KwK_{\mathrm w} of KK in XwX_{\mathrm w} is the subset in (1), equipped with its subspace topology: Kw={z∈Xw:ι−1(z)∈K}.K_{\mathrm w} =\{z\in X_{\mathrm w}:\iota^{-1}(z)\in K\}.(1)✦ The norm-to-weak identification ιK\iota_K is the bijection from KK, equipped with its norm topology, to KwK_{\mathrm w} given by (2): ιK:K→Kw,ιK(x)=ι(x),ιK−1(z)=ι−1(z).\iota_K\colon K\longrightarrow K_{\mathrm w},\qquad \iota_K(x)=\iota(x),\qquad \iota_K^{-1}(z)=\iota^{-1}(z).(2)✦

A weakly closed subset KK of XX is a set whose image under ι\iota is closed in XwX_{\mathrm w}, as in (3): K is weakly closed⇔ι(K) is closed in Xw.K\text{ is weakly closed} \quad\Longleftrightarrow\quad \iota(K)\text{ is closed in }X_{\mathrm w}.(3)✦ A subset KK of XX is weakly compact when its image under ι\iota is compact in XwX_{\mathrm w}, as in (4): K is weakly compact⇔ι(K) is compact in Xw.K\text{ is weakly compact} \quad\Longleftrightarrow\quad \iota(K)\text{ is compact in }X_{\mathrm w}.(4)✦

The closed difference span YY of KK is the norm closure of the real linear span of K−KK-K, defined in (5): Y=span⁡ℝ(K−K)¯∥⋅∥.Y=\overline{\operatorname{span}_{\mathbb R}(K-K)}^{\|\cdot\|}.(5)✦ The difference ball BYB_Y is the whole closed unit ball of YY, viewed as the subset of XX in (6): BY={x∈X:x∈Y and ∥x∥≤1}.B_Y=\{x\in X:x\in Y\text{ and }\|x\|\leq 1\}.(6)✦

A subset ss of XX is invariant under TT if TT sends every point of s∩Ks\cap K into ss, as in (7): s is invariant under T⇔T(s∩K)⊆s.s\text{ is invariant under }T \quad\Longleftrightarrow\quad T(s\cap K)\subseteq s.(7)✦ The subset KK of XX is minimal invariant under TT if it has no proper nonempty norm-closed convex invariant subset, as in (8): K is minimal invariant under T⇔every nonempty norm-closed convex subset s⊆Kthat is invariant under T equals K.\begin{gathered} K\text{ is minimal invariant under }T \quad\Longleftrightarrow\quad \text{every nonempty norm-closed convex subset }s\subseteq K\\ \text{that is invariant under }T\text{ equals }K. \end{gathered}(8)✦

What Is Checked

These definitions are stated for a real normed space XX, an arbitrary subset K⊆XK\subseteq X and a selfmap T:K→KT\colon K\to K.