OctoInformBeta

OpenAI preprint, September 2026 · Appendix A. The fixed-point ingredients

Lemma A.2 [Fixed points on finite probability simplices]: The coordinates may have any nonempty finite index set, including a singleton

Continuous self-maps of finite probability simplices have 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 ι\iota be a finite index set. A probability weight vector aa on ι\iota is a function a:ι→ℝa\colon\iota\to\mathbb{R} whose coordinates are nonnegative and sum to one.✦ The probability simplex Δ(ι)\Delta(\iota) is the subset of ℝι\mathbb{R}^{\iota} consisting of probability weight vectors on ι\iota, with the inherited topology: Δ(ι)={a∈ℝι:ai≥0 for every i∈ι,∑i∈ιai=1}.\Delta(\iota) = \left\{a\in\mathbb{R}^{\iota}: a_i\geq 0\text{ for every }i\in\iota,\quad \sum_{i\in\iota}a_i=1\right\}.(1)✦ For a nonnegative integer mm, the standard simplex Δm\Delta_m is the probability simplex in (1) with index set {0,…,m}\{0,\ldots,m\}.

The Claim

Theorem 1. Let mm be a nonnegative integer and let P:Δm→ΔmP\colon\Delta_m\to\Delta_m be continuous. Then ∃a∈ΔmP(a)=a.\exists a\in\Delta_m\quad P(a)=a.(2)✦

Theorem 2. Let ι\iota be a nonempty finite index set and let P:Δ(ι)→Δ(ι)P\colon\Delta(\iota)\to\Delta(\iota) be continuous. Then ∃a∈Δ(ι)P(a)=a.\exists a\in\Delta(\iota)\quad P(a)=a.(3)✦

What Is Checked

The statements (2) and (3) concern continuous self-maps of the probability simplices in (1), with a nonempty finite coordinate index set.