Alphabeta Math
LemmaStatement: AI-adaptedProof: AI-generatedPipeline-generatedjudge pass (gpt-6-luna)
How statement and proof provenance work

The first chip identifies the source of the statement or construction; the second identifies the source of its local proof or verification.

  • Literature-sourced: the exact statement appears in a cited source; only wording and notation differ.
  • AI-adapted: a semantically identical restatement of literature-sourced material, modulo indexing, notation, and boundary cases adopted by the library.
  • AI-generated: a genuinely novel statement formulated by AI, with no source for the claim itself.

These labels describe origin, not correctness: citations and verification chips remain separate evidence.

Local Gram positivity and loss smoothness give confined gradient descent

Statement

Let u:Rp→Rn be C2, L=12∥u−y∥2, and r0=u(θ0)−y. On the closed ball B(θ0,R) suppose a block J∗ of the output Jacobian satisfies J∗J∗T⪰(λ/2)In and ∥∇2L∥≤S, with λ,S>0. If R>22∥r0∥/λ and 0<η≤1/S, every fixed-step gradient-descent iterate stays in the ball,

L(θk)≤(1−ηλ/2)kL(θ0),∑k≥0∥θk+1−θk∥≤22∥r0∥/λ.

The assertion is trivial for r0=0; otherwise the hypotheses imply 0<ηλ/2≤1.

Facts & Assumptions

Given: The output map, quadratic loss, ball, spectral and Hessian bounds, and step size.

[F1]

Gradient descent uses θ+=θ−η∇L(θ). Gradient descent

[F2]

For a scalar C2 function on a segment, the first-order Taylor remainder is the integral of its second derivative against the linear remainder kernel. Taylor's theorem with an integral remainder

[F3]

The Hessian norm is the induced operator norm on the finite-dimensional Euclidean parameter space. The operator norm is zero on the zero domain and otherwise is max_{||v||=1} ||Tv||

[F4]

Euclidean Cauchy–Schwarz gives ∣aTb∣≤∥a∥2∥b∥2 for all real vectors of the same finite dimension. Cauchy-Schwarz ∣⟨x,y⟩∣≤∥x∥2∥y∥2 with its equality case, the triangle inequality for ∥⋅∥2, the parallelogram law and polarisation

Proof

1.1F1F2F3F4givenalgebra

Write g=∇L=JTr. Since JJT⪰J∗J∗T, ∥g∥2=rTJJTr≥(λ/2)∥r∥2=λL in the ball. For any step θ+=θ−hg, 0<h≤η, whose segment stays in the ball, put ψ(s)=L(θ−sg) for 0≤s≤h. Then ψ′(0)=−∥g∥2 and ψ′′(s)=gTHsg, where Hs=∇2L(θ−sg). By Euclidean Cauchy–Schwarz [F4] and the induced norm [F3], ∣gTHsg∣≤∥g∥∥Hsg∥≤∥g∥∥Hs∥∥g∥≤S∥g∥2; for g=0 the same bound is immediate. By [F2], L+=L−h∥g∥2+∫0h(h−s)ψ′′(s) ds≤L−h(1−Sh/2)∥g∥2≤(1−hλ/2)L. If L>0, nonnegativity of L+ gives hλ/2≤1.

2.1step 1.1algebra

For such a segment with L>0, L−L+=(L−L+)/(L+L+)≥h∥g∥2/(4L)≥(λ/4)h∥g∥. If L=0, r=0 and hence g=0, so the step is stationary. Telescoping any finite sequence of admissible full or truncated steps gives total parameter path length at most 4L0/λ=22∥r0∥/λ.

3.1step 1.1step 2.1algebra∎

Suppose there is a first proposed full step exiting the ball. Truncate that step at the first intersection with its boundary; this is a segment of length h∥g∥ with h≤η, wholly in the ball. Step 2.1 applies to all previous steps and this truncated step, giving total path length strictly less than R. Reaching the boundary from θ0 requires length at least R, a contradiction. Thus no exit occurs. Step 1.1 applied at every full step gives the loss rate, and taking the supremum of finite partial path lengths in step 2.1 gives the asserted series bound.

Depends on

Used by

Dependency tree · two levels

28 results within two dependency steps of this one, each drawn at its shortest distance from it. An arrow runs from a result to what uses it, so the chart reads left to right and ends at this result, which carries a heavier outline. Every node is a link to that result. Click elsewhere on the chart to enlarge it.

Sources