OctoInformBeta

OpenAI preprint, September 2026 · §2. An obstruction to deterministic selection

§2 (setting): the five-component mixture: A finite mixture allows nonnegative extended weights

Combining a measure with its branch images.

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 ν\nu be a central measure on a measurable space AA, and let Hi:A→AH_i\colon A\to A, i∈Ii\in I, be a finite family of branch maps. The notation (Hi)#ν(H_i)_\#\nu denotes the pushforward of ν\nu under HiH_i. Let aa and bib_i, i∈Ii\in I, be weights in [0,+∞][0,+\infty].

The Definition

Definition 1. The weighted pushforward mixture μ\mu of a measure ν\nu on a measurable space AA with a finite family of branch maps Hi:A→AH_i\colon A\to A, i∈Ii\in I, and weights aa and bib_i, i∈Ii\in I, in [0,+∞][0,+\infty] is the measure on AA given by (1): μ=aν+∑i∈Ibi(Hi)#ν.\mu = a\,\nu + \sum_{i\in I} b_i\,(H_i)_\#\nu.(1)✦

In the paper, the result reads: given a probability ν\nu concentrated on a set BB and four branch maps H1,…,H4H_1,\ldots,H_4, put μ=ν/3+(1/6)∑j=14(Hj)#ν\mu=\nu/3+(1/6)\sum_{j=1}^{4}(H_j)_\#\nu.

What Is Checked

The definition (1) is stated over a measure on an arbitrary measurable space, a finite family of maps from that space to itself, and nonnegative extended weights.