OctoInformBeta

OpenAI preprint, September 2026 · §3. From affine labels to persistent trajectories

Proposition 3.3: For finite networks with positive constant reaction rates and any positive initial state

Weak reversibility gives positive trapping polytopes.

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 question concerns trapping regions for mass-action dynamics in the positive orthant, without a boundedness hypothesis on the stoichiometric class.

A reaction network NN on dd species consists of a finite complex set C⊆ℕdC\subseteq\mathbb N^d and a finite reaction set R⊆C×CR\subseteq C\times C whose pairs have distinct entries.✦ A reaction y→zy\to z of the network NN is a pair (y,z)∈R(y,z)\in R. The network NN is weakly reversible if every reaction y→zy\to z has a directed path of one or more reactions from zz to yy.✦

A positive state x∈ℝdx\in\mathbb R^d is a vector with xi>0x_i>0 for every species index ii.✦ Assign a positive constant rate κy→z\kappa_{y\to z} to each reaction. For a real state x∈ℝdx\in\mathbb R^d and a complex y∈Cy\in C, the mass-action monomial xyx^y and the mass-action vector field f(x)f(x) are given by (1). xy=∏i=1dxiyi,f(x)=∑(y,z)∈Rκy→zxy(z−y).x^y=\prod_{i=1}^{d}x_i^{y_i}, \qquad f(x)=\sum_{(y,z)\in R}\kappa_{y\to z}x^y(z-y).(1)

Solutions and Trapping Regions

For T≥0T\ge0, a forward solution from x0∈ℝdx_0\in\mathbb R^d on [0,T][0,T] is a trajectory x:ℝ→ℝdx:\mathbb R\to\mathbb R^d that is continuous on [0,T][0,T] and has the initial value and ordinary two-sided derivatives in (2). x(0)=x0,x′(t)=f(x(t))(0<t<T).x(0)=x_0, \qquad x'(t)=f(x(t))\quad(0<t<T).(2)✦

A global forward solution from x0∈ℝdx_0\in\mathbb R^d is a trajectory x:ℝ→ℝdx:\mathbb R\to\mathbb R^d with the initial value and ordinary two-sided derivatives in (3). x(0)=x0,x′(t)=f(x(t))(t≥0).x(0)=x_0, \qquad x'(t)=f(x(t))\quad(t\ge0).(3)✦

A forward-invariant set K⊆ℝdK\subseteq\mathbb R^d for ff is a set such that, for each z∈Kz\in K, there is a global forward solution from zz, and every forward solution from zz on any interval [0,T][0,T], T≥0T\ge0, stays in KK throughout that interval and agrees there with a global forward solution.✦

For real vectors a,x∈ℝda,x\in\mathbb R^d, their dot product is a⋅x=∑i=1daixia\cdot x=\sum_{i=1}^d a_i x_i.✦ A finite polyhedron K⊆ℝdK\subseteq\mathbb R^d is an intersection of finitely many closed half-spaces, as in (4), where n≥0n\ge0, aj∈ℝda_j\in\mathbb R^d and bj∈ℝb_j\in\mathbb R. K={x∈ℝd:bj≤aj⋅x for 1≤j≤n}.K=\{x\in\mathbb R^d:b_j\le a_j\cdot x\text{ for }1\le j\le n\}.(4)✦

The Claim

Theorem 1. Let d>0d>0 and let NN be a finite weakly reversible reaction network on dd species. Suppose all constant reaction rates κy→z\kappa_{y\to z} are positive, and let x0∈(0,∞)dx_0\in(0,\infty)^d. There is a compact convex finite polyhedron K⊆ℝdK\subseteq\mathbb R^d satisfying (5). x0∈K⊆(0,∞)d,K is forward invariant for f.x_0\in K\subseteq(0,\infty)^d, \qquad K\text{ is forward invariant for }f.(5)✦

In the paper, the result reads: for every finite weakly reversible mass-action network with positive constant rates and every strictly positive initial state x0x_0, there is a compact convex polytope K⊆(0,∞)dK\subseteq(0,\infty)^d containing x0x_0 and forward invariant for ẋ=f(x)\dot x=f(x), and every solution starting in KK exists for all forward time and stays in KK.

What Is Checked

The claim is stated over a finite weakly reversible network on d>0d>0 species with positive constant reaction rates, and an initial state in (0,∞)d(0,\infty)^d.