OctoInformBeta

OpenAI preprint, September 2026 · §3. Attaining the maximal defect

§3 (setting): signed defect, its bound and its supremum: The absolute supremum is invariant under inversion, with separate estimates near each endpoint

Sphere isometries have bounded radial defect.

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 XX and YY be real normed spaces. The unit spheres SXS_X of XX and SYS_Y of YY are the subsets given in (1), equipped with the inherited distances: SX={x∈X:∥x∥=1},SY={y∈Y:∥y∥=1}.S_X=\{x\in X:\|x\|=1\},\qquad S_Y=\{y\in Y:\|y\|=1\}.(1)✦ Let f:SX→SYf:S_X\to S_Y be a surjective isometry. For x,y∈SXx,y\in S_X and 0≤q≤10\le q\le1, the quantities to be compared are ∥x−qy∥\|x-qy\| and ∥f(x)−qf(y)∥\|f(x)-qf(y)\|: each is the distance from a unit vector to a radial contraction of another unit vector.

The Defect

Definition 1. The signed defect Dq(x,y)D_q(x,y) of ff, for q∈ℝq\in\mathbb R and x,y∈SXx,y\in S_X, is the real number defined by (2): Dq(x,y)=∥f(x)−qf(y)∥−∥x−qy∥.D_q(x,y)=\|f(x)-qf(y)\|-\|x-qy\|.(2)✦ A uniform defect bound bb for ff is a real number satisfying (3): |Dq(x,y)|≤b(x,y∈SX,0≤q≤1).|D_q(x,y)|\le b \qquad(x,y\in S_X,\ 0\le q\le1).(3)✦ The absolute defect set 𝒜f\mathcal A_f of ff is the subset of ℝ\mathbb R given by (4): 𝒜f={|Dq(x,y)|:x,y∈SX,0≤q≤1}.\mathcal A_f=\{|D_q(x,y)|:x,y\in S_X,\ 0\le q\le1\}.(4)✦ The maximal defect M=M(f)M=M(f) of ff is the supremum in (5): M=M(f)=sup⁡𝒜f=sup⁡{|Dq(x,y)|:x,y∈SX,0≤q≤1}.M=M(f)=\sup\mathcal A_f =\sup\{|D_q(x,y)|:x,y\in S_X,\ 0\le q\le1\}.(5)✦

Write Dqf−1D^{\,f^{-1}}_q for the signed defect of the inverse isometry f−1:SY→SXf^{-1}:S_Y\to S_X, and M(f−1)M(f^{-1}) for its maximal defect.

The Claim

Theorem 1. Let X,YX,Y be real normed spaces and f:SX→SYf:S_X\to S_Y a surjective isometry. For q∈ℝq\in\mathbb R and x,y∈SXx,y\in S_X, the signed defects satisfy (6); the maximal defects satisfy (7): Dqf−1(f(x),f(y))=−Dq(x,y).D^{\,f^{-1}}_q(f(x),f(y))=-D_q(x,y).(6)✦ M(f−1)=M(f).M(f^{-1})=M(f).(7)✦ For 0≤q≤10\le q\le1 and x,y∈SXx,y\in S_X, the endpoint estimates are (8): |Dq(x,y)|≤2q,|Dq(x,y)|≤2(1−q).|D_q(x,y)|\le2q,\qquad |D_q(x,y)|\le2(1-q).(8)✦ If SXS_X is nonempty, the maximal defect satisfies (9) and (10): 0≤M.0\le M.(9)✦ M≤1.M\le1.(10)✦

In the paper, the result reads: for a surjective sphere isometry f:SX→SYf:S_X\to S_Y, put Dq(x,y)=∥f(x)−qf(y)∥−∥x−qy∥D_q(x,y)=\|f(x)-qf(y)\|-\|x-qy\| and M=sup⁡{|Dq(x,y)|:x,y∈SX,0≤q≤1}M=\sup\{|D_q(x,y)|:x,y\in S_X,\ 0\le q\le1\}; the inverse has the negative signed defect and the same absolute supremum, D0(x,y)=D1(x,y)=0D_0(x,y)=D_1(x,y)=0, |Dq(x,y)|≤2min⁡{q,1−q}|D_q(x,y)|\le2\min\{q,1-q\} for x,y∈SXx,y\in S_X and 0≤q≤10\le q\le1, and 0≤M≤10\le M\le1.

What Is Checked

The library’s statements concern surjective isometries between unit spheres of real normed spaces, with a nonempty source sphere in (9) and (10).