Spectral Gradient Quadratic Convergence — Minimization with Minimizer-Anchored Hessian Regularity
Descriptions may be stale. Consult the linked Lean sources.
Minimization with Minimizer-Anchored Hessian Regularity · Lean proof
Minimization
with Minimizer-Anchored Hessian Regularity
We consider minimizing a differentiable function on a nonzero
finite-dimensional real Hilbert space . The function is -strongly convex with an -Lipschitz, differentiable gradient,
where . A given point
lies within distance of a global minimizer , and the Hessian satisfies for every , where and the norms are induced by the
inner product. The oracle returns at . For , the goal is an output
satisfying
almost surely under , the law of
the algorithm’s randomness.
The optimization problem is The algorithm receives , , , , , , and . We write for the induced
operator norm; each Hessian is a linear operator from to .
Our assumptions:
Assumption 1 (Strong convexity). For every ,
Assumption 2 (Differentiability and smoothness).
Both and are Fréchet differentiable
everywhere on , and
Assumption 3 (Initial distance and Hessian
regularity). There exists such that and the Hessian obeys the minimizer-anchored
bound
Queries may be made at any point of . After the allotted oracle calls, the
output must satisfy
Proved result
OracleTemplate.Convergence LongTails.Setups.MinimizerRegularHessian.spec OracleTemplate.SpectralGradient.budget
Relevant previous results
No qualifying prior reference verified; this is not a proof of novelty.