OctoInformBeta

OpenAI preprint, September 2026 · 2. An entropy inequality with one zero block · Section 2

Lemma 2.5: For square-root determinants in the closed interval [0,1/2][0,1/2]

Convexity and monotonicity of the Wootters function.

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

Qubit entropy can be expressed as a scalar function of the determinant. Binary entropy hh is the entropy in nats of a binary probability distribution with probabilities qq and 1−q1-q, defined for q∈[0,1]q\in[0,1] by (1), with 0ln⁡0=00\ln 0=0: h(q)=−qln⁡q−(1−q)ln⁡(1−q).h(q)=-q\ln q-(1-q)\ln(1-q).(1)

The qubit entropy function gg at determinant u∈[0,1/4]u\in[0,1/4] is given by (2): g(u)=h(1+1−4u2).g(u)=h\!\left(\frac{1+\sqrt{1-4u}}{2}\right).(2)✦ The Wootters entropy function ww is the real scalar function that assigns to each s∈ℝs\in\mathbb R the value of gg at s2s^2. w(s)=g(s2),0≤s≤12.w(s)=g(s^2),\qquad 0\le s\le\frac12.(3)✦

The Claim

Theorem 1. Let ss and tt belong to the closed interval [0,1/2][0,1/2]. For every λ∈[0,1]\lambda\in[0,1], the convexity bound (4) holds: w(λs+(1−λ)t)≤λw(s)+(1−λ)w(t).w\!\left(\lambda s+(1-\lambda)t\right) \le \lambda w(s)+(1-\lambda)w(t).(4)✦ If s≤ts\le t, the monotonicity bound is (5): w(s)≤w(t).w(s)\le w(t).(5)✦

What Is Checked

The claims concern the real scalar function s↦g(s2)s\mapsto g(s^2), defined through binary entropy, on the closed interval [0,1/2][0,1/2], including both endpoints.