OctoInformBeta

OpenAI preprint, September 2026 · §5. A dual space with coded norming functionals

§5 (setting): rational normers, coded paths and the completed space: Crops and coded finite paths generate pure and mixed families and the coordinate predual

Rational normers and their completed space.

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

The construction assigns a norm to finitely supported real functions on Γ=ω1\Gamma=\omega_1 by a family of finitely supported rational arrays. Pure operations act within a fixed color, while mixed operations also impose succession of colors. The completed real space is EE✦; the closed span of its coordinate functionals is X0X_0.✦

Fix a bijection π:Γ→ℕ×Γ\pi:\Gamma\to\mathbb N\times\Gamma.✦ The color of α\alpha is c(α)=π(α)1c(\alpha)=\pi(\alpha)_1, and Ck=c−1({k})C_k=c^{-1}(\{k\}) is an unbounded set of cardinality ℵ1\aleph_1.✦ Write 𝒜=c00(Γ,ℚ)\mathcal A=c_{00}(\Gamma,\mathbb Q) for the rational arrays, δα\delta_\alpha for the array equal to one at α\alpha and zero elsewhere, and u|Du|_D for the restriction of u∈𝒜u\in\mathcal A to D⊆ΓD\subseteq\Gamma, extended by zero.

The operation families are 𝗉k\mathsf p_k, for k∈ℕk\in\mathbb N, and 𝗆\mathsf m.✦ Their parameter indices are s(𝗉k)=k+2s(\mathsf p_k)=k+2 and s(𝗆)=2s(\mathsf m)=2✦, and their exponent is qf=(s(f)+1)/s(f)q_f=(s(f)+1)/s(f).✦ For arrays u,vu,v, write u≺fvu\prec_f v when every α∈supp⁡u\alpha\in\operatorname{supp}u and β∈supp⁡v\beta\in\operatorname{supp}v satisfy α<β\alpha<\beta, with the additional condition c(α)<c(β)c(\alpha)<c(\beta) for f=𝗆f=\mathsf m.✦

Parameters and Prefix Codes

The depth sequence is dn=n+11d_n=n+11, with zero-based index nn.✦ For a natural parameter ss, the exponent scale ts,nt_{s,n} is the first coordinate of a recursive state (ts,n,bs,n)(t_{s,n},b_{s,n}).✦ Its recursion is (ts,0,bs,0)=(d02+2d0+2,0),bs,n+1=bs,n+2⋅2s(s+1)ts,n,ts,n+1=ts,n+(bs,n+1+1)(dn+1+1)+1+dn+1+dn+12.\begin{aligned} (t_{s,0},b_{s,0})&=(d_0^2+2d_0+2,0),\\ b_{s,n+1}&=b_{s,n}+2\cdot2^{s(s+1)t_{s,n}},\\ t_{s,n+1}&=t_{s,n}+(b_{s,n+1}+1)(d_{n+1}+1) +1+d_{n+1}+d_{n+1}^2. \end{aligned}(1)✦ The block length and normalization factor at weight j≥1j\geq1 use parameter index j−1j-1 in (1): Ljf=2s(s+1)ts,j−1,mjf=2(s−1)(s+1)ts,j−1+dj−1,s=s(f).L_j^f=2^{s(s+1)t_{s,j-1}},\qquad m_j^f=2^{(s-1)(s+1)t_{s,j-1}+d_{j-1}},\qquad s=s(f).(2)

A predecessor closure family (Gn(a))a<β,n∈ℕ(G_n(a))_{a<\beta,\,n\in\mathbb N} at β\beta is a family of finite sets indexed by a<βa<\beta and n∈ℕn\in\mathbb N.✦ A coherent predecessor closure family at β\beta is a family of finite sets Gn(a)G_n(a), for a<βa<\beta and n∈ℕn\in\mathbb N, whose sequences increase, contain their endpoints, exhaust the intervals up to their endpoints, and satisfy a∈Gn(a)⊆[0,a],⋃nGn(a)=[0,a],b∈Gn(a)⇒Gn(b)=Gn(a)∩[0,b].a\in G_n(a)\subseteq[0,a],\qquad \bigcup_nG_n(a)=[0,a],\qquad b\in G_n(a)\ \Longrightarrow\ G_n(b)=G_n(a)\cap[0,b].(3) A coherent endpoint extension of a family satisfying (3) is an increasing sequence of finite sets Hn(β)H_n(\beta) containing β\beta, contained in [0,β][0,\beta], and exhausting [0,β][0,\beta], with Gn(a)=Hn(β)∩[0,a]G_n(a)=H_n(\beta)\cap[0,a] whenever a<βa<\beta belongs to Hn(β)H_n(\beta).✦ The endpoint construction extends predecessor families; transfinite recursion gives the coherent finite closure system Hn(β)H_n(\beta). Its positive-indexed closures are denoted by Fp(β)=Hmax⁡{p−1,0}(β)F_p(\beta)=H_{\max\{p-1,0\}}(\beta). For α,β∈Γ\alpha,\beta\in\Gamma, the closure entry rank ρ(α,β)∈ℕ\rho(\alpha,\beta)\in\mathbb N is the first positive index pp for which α∈Fp(β)\alpha\in F_p(\beta) when α≤β\alpha\leq\beta, and is zero otherwise.

A raw infinite path is a sequence P=(ui,ji,pi)i≥0P=(u_i,j_i,p_i)_{i\geq0} of nonzero arrays ui∈𝒜u_i\in\mathcal A, weights ji∈ℕj_i\in\mathbb N, and metadata pi∈ℕp_i\in\mathbb N.✦ For a raw infinite path PP, the prefix support SnS_n is a finite subset of Γ\Gamma, the endpoint βn\beta_n is its largest element, and the prefix closure DnD_n is the finite subset of Γ\Gamma given by Sn=⋃i=0nsupp⁡ui,βn=max⁡Sn,Dn=Fpn(βn).S_n=\bigcup_{i=0}^{n}\operatorname{supp}u_i,\qquad \beta_n=\max S_n,\qquad D_n=F_{p_n}(\beta_n).(4) For a finite set D⊆ΓD\subseteq\Gamma, the position of α\alpha in DD is pos⁡D(α)=|D∩[0,α)|\operatorname{pos}_D(\alpha)=|D\cap[0,\alpha)|.✦ For an array u∈𝒜u\in\mathcal A and a finite set D⊆ΓD\subseteq\Gamma, the positional array code ac⁡D(u)\operatorname{ac}_D(u) is the finite set of pairs ac⁡D(u)={(pos⁡D(α),u(α)):α∈D}\operatorname{ac}_D(u)=\{(\operatorname{pos}_D(\alpha),u(\alpha)):\alpha\in D\}.✦ A finite path code cc is a finite list of triples consisting of a weight, metadata, and a finite set of pairs from ℕ×ℚ\mathbb N\times\mathbb Q.✦ The prefix code of PP is the list code⁡n(P)=[(ji,pi,ac⁡Dn(ui))]i=0n.\operatorname{code}_n(P) =\bigl[(j_i,p_i,\operatorname{ac}_{D_n}(u_i))\bigr]_{i=0}^{n}.(5)✦ For a code cc, let M(c)M(c) be its largest metadata entry, with M(⌀)=0M(\varnothing)=0.✦ Fix an injective numbering ν\nu of these codes. The coding map σ\sigma from finite path codes to ℕ\mathbb N is the map defined by σ(c)=⟨ν(c),M(c)+1⟩\sigma(c)=\langle\nu(c),M(c)+1\rangle, where the brackets denote the natural-number pairing function.✦ The indices in (4) and (5) begin at zero.

The Definition

Definition 1. A convex crop AA is a pair of order-convex sets, one in Γ\Gamma and one in ℕ\mathbb N: A=(I,J),I⊆Γ order-convex,J⊆ℕ order-convex.A=(I,J),\qquad I\subseteq\Gamma\text{ order-convex},\qquad J\subseteq\mathbb N\text{ order-convex}.(6)✦ The support set of the crop in (6) is Af=IA_f=I for a pure family and A𝗆=I∩c−1(J)A_{\mathsf m}=I\cap c^{-1}(J) for the mixed family.✦

A Type I expression ee for ff is a positive weight jj and a nonempty finite sequence of nonzero arrays satisfying e=(j;v0,…,vℓ−1),1≤ℓ≤Ljf,vh≠0,vh≺fvi(h<i<ℓ).e=(j;v_0,\ldots,v_{\ell-1}),\qquad 1\leq\ell\leq L_j^f,\qquad v_h\ne0,\quad v_h\prec_f v_i\quad(h<i<\ell).(7)✦ The value V(e)V(e) of a Type I expression ee for ff is the normalized rational array V(e)=1mjf∑h<ℓvh.V(e)=\frac1{m_j^f}\sum_{h<\ell}v_h.(8)

An admissible prefix of length ℓ\ell of a raw path PP is a finite initial segment of PP of length ℓ\ell that satisfies supp⁡ui<supp⁡uh,pi<ph(i<h<ℓ),1≤ji≤pi(i<ℓ),ρ(α,β)≤pn(n<ℓ,α,β∈Sn,α≤β),jn+1=σ(code⁡n(P))(n+1<ℓ).\begin{aligned} &\operatorname{supp}u_i<\operatorname{supp}u_h, \quad p_i<p_h &&(i<h<\ell),\\ &1\leq j_i\leq p_i &&(i<\ell),\\ &\rho(\alpha,\beta)\leq p_n &&(n<\ell,\ \alpha,\beta\in S_n,\ \alpha\leq\beta),\\ &j_{n+1}=\sigma(\operatorname{code}_n(P)) &&(n+1<\ell). \end{aligned}(9)✦ The support inequality in (9) means that each point of the first support precedes each point of the second.

A finite admissible path for ff is a raw path PP, a positive length ℓ\ell, and Type I expressions eie_i with values given by (8) for its first ℓ\ell pieces, subject to $$\begin{gathered} \ell\geq1,\qquad P\text{ satisfies }\eqref{eq:claim3},\\ V(e_i)=u_i,\quad \operatorname{weight}(e_i)=j_i\quad(i<\ell),\qquad u_i\prec_f u_h\quad(i<h<\ell). \end{gathered}$$(10)✦ The pieces uiu_i in (10) are the original uncropped rational arrays. For a finite path and a crop, the active weights form the finite set Wf(P,A)={ji:i<ℓ,ui|Af≠0}.W_f(P,A)=\{j_i:i<\ell,\ u_i|_{A_f}\ne0\}.(11)✦

A Type II expression η\eta for ff is a finite family of finite admissible paths PbP_b, crops AbA_b, and rational coefficients aba_b such that ∑b<N|ab|qf≤1,Wf(Pb,Ab)∩Wf(Pc,Ac)=⌀(b≠c).\sum_{b<N}|a_b|^{q_f}\leq1,\qquad W_f(P_b,A_b)\cap W_f(P_c,A_c)=\varnothing\quad(b\ne c).(12)✦ The active sets in (12) are those of (11); NN may be zero. Denote the Type II value by V(η)V(\eta).

For a base B⊆𝒜B\subseteq\mathcal A, the construction stages Kn(B,f)K_n(B,f) are $$\begin{aligned} K_0(B,f)&=B,\\ K_{n+1}(B,f)&=K_n(B,f)\\ &\quad\cup\{V(e):e\text{ satisfies }\eqref{eq:claim2},\ v_h\in K_n(B,f)\text{ for every child }v_h\}\\ &\quad\cup\{V(\eta):\eta\text{ satisfies }\eqref{eq:claim5},\ \text{every child of every path expression belongs to }K_n(B,f)\}. \end{aligned}$$(13)✦ The stage index in (13) begins at zero. The pure norming family of color kk is 𝒦k=⋃n≥0Kn(Bk,𝗉k),Bk={0}∪{δα,−δα:α∈Ck}.\mathcal K_k=\bigcup_{n\geq0}K_n(B_k,\mathsf p_k),\qquad B_k=\{0\}\cup\{\delta_\alpha,-\delta_\alpha:\alpha\in C_k\}.(14)✦ The full norming family is the mixed closure of the pure families in (14): 𝒦=⋃n≥0Kn(⋃k≥0𝒦k,𝗆).\mathcal K=\bigcup_{n\geq0}K_n\left(\bigcup_{k\geq0}\mathcal K_k,\mathsf m\right).(15)✦

A norming set on Γ\Gamma is a set K⊆𝒜K\subseteq\mathcal A that contains every δα\delta_\alpha and satisfies |f(α)|≤1|f(\alpha)|\leq1 for every f∈Kf\in K and α∈Γ\alpha\in\Gamma.✦ The norming set for the completed space is K=𝒦.K=\mathcal K.(16)✦ For this KK from (16), let ℓ∞(K)\ell_\infty(K) be the real bounded functions on KK.✦ The coordinate evaluation vector vα∈ℓ∞(K)v_\alpha\in\ell_\infty(K) is vα(f)=f(α)v_\alpha(f)=f(\alpha).✦ The finite-support inclusion is the real linear map Jx=∑αx(α)vαJx=\sum_\alpha x(\alpha)v_\alpha from c00(Γ,ℝ)c_{00}(\Gamma,\mathbb R) into ℓ∞(K)\ell_\infty(K).✦ The norming completion EKE_K of the norming set KK is the closed span of the coordinate evaluation vectors in ℓ∞(K)\ell_\infty(K), given by E=EK=J[c00(Γ,ℝ)]¯ℓ∞(K).E=E_K=\overline{J[c_{00}(\Gamma,\mathbb R)]}^{\,\ell_\infty(K)}.(17)✦ For f∈Kf\in K, the norming evaluation functional on EE in (17) is φf(z)=z(f)\varphi_f(z)=z(f).✦ The coordinate functionals are εα=φδα∈E*\varepsilon_\alpha=\varphi_{\delta_\alpha}\in E^*.✦ The coordinate predual is X0=span⁡ℝ{εα:α∈Γ}¯E*.X_0=\overline{\operatorname{span}_{\mathbb R} \{\varepsilon_\alpha:\alpha\in\Gamma\}}^{\,E^*}.(18)✦ The closure in (18) is in the dual norm.

What Is Checked

The definitions are stated over ω1\omega_1, its fixed color partition and coherent finite closure system, the parameter sequences, and the rational path coding map. Every raw prefix support SnS_n in (4) is nonempty.✦