OctoInformBeta

OpenAI preprint, September 2026 · §2. Affine labels with a prescribed tolerance

§2 (setting): Power coordinates and exponent cubes at small positive scales

Affine branches and their active labels.

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

For a mass-action system with dd species, consider a positive concentration state x∈(0,∞)dx\in(0,\infty)^d and a scale 0<h<10<h<1. The affine construction compares branches at xx while their offsets vary with the scale; exponent coordinates describe the states at which these comparisons are made.

The Claim

Definition 1. An affine label LL on ℝd\mathbb R^d is a pair (r,c)(r,c) consisting of a slope r∈ℝdr\in\mathbb R^d and an offset function c:ℝ→ℝc:\mathbb R\to\mathbb R.✦ The affine branch value vL(h,x)v_L(h,x) of an affine label L=(r,c)L=(r,c) on ℝd\mathbb R^d at a real scale hh and a vector x∈ℝdx\in\mathbb R^d is the number given by (1): vL(h,x)=r⋅x+c(h)=∑i=1drixi+c(h).v_L(h,x)=r\cdot x+c(h) =\sum_{i=1}^{d}r_i x_i+c(h).(1)✦

For real numbers a,ba,b and a nonnegative integer mm, the exponent cube [a,b]m[a,b]^m is the set of vectors specified by (2): [a,b]m={p∈ℝm:a≤pi≤b for every i=1,…,m}.[a,b]^m =\{p\in\mathbb R^m: a\leq p_i\leq b\text{ for every }i=1,\ldots,m\}.(2)✦ The power coordinate vector Xh(p)X_h(p) associated with an exponent vector p∈ℝmp\in\mathbb R^m and a scale hh is defined by (3): Xh(p)=(hpi)i=1m.X_h(p)=(h^{p_i})_{i=1}^{m}.(3)✦

For a finite family Λ\Lambda of affine labels on ℝd\mathbb R^d, an active label at (h,x)(h,x) is a label LL satisfying (4): Active⁡Λ(L;h,x)⇔L∈Λ and vL(h,x)≤vJ(h,x) for every J∈Λ.\operatorname{Active}_{\Lambda}(L;h,x) \quad\Longleftrightarrow\quad L\in\Lambda\ \text{ and }\ v_L(h,x)\leq v_J(h,x)\text{ for every }J\in\Lambda.(4)✦

Theorem 1. Let Λ\Lambda be a nonempty finite family of affine labels on ℝd\mathbb R^d, let h∈ℝh\in\mathbb R, and let x∈ℝdx\in\mathbb R^d. For any affine label LL, the minimum characterization (5) holds, including all tied minimizers: Active⁡Λ(L;h,x)⇔L∈Λ and vL(h,x)=minJ∈ΛvJ(h,x).\operatorname{Active}_{\Lambda}(L;h,x) \quad\Longleftrightarrow\quad L\in\Lambda\ \text{ and }\ v_L(h,x)=\min_{J\in\Lambda}v_J(h,x).(5)✦

What Is Checked

The definitions allow real scales and real coordinate vectors; the characterization (5) is stated for a nonempty finite family of affine labels with arbitrary slope vectors and offset functions.