arXiv:2610.23695
Recursive Operator Localization and Spectral Gap Descent via the Geometry of Local Spectral Expansion
Ryu Hayakawa
Abstract
We introduce Recursive Operator Localization (ROL), a framework for constructing hierarchies of local energy problems linked by linear maps between Hilbert spaces. To address spectral-gap problems, including those arising in quantum systems, we study a realization of ROL by nested subspaces indexed by a weighted simplicial complex. Using spectral gaps of auxiliary classical random walks on links, we propagate geometric estimates upward and use them to turn local spectral gaps at the bottom of the hierarchy into a spectral-gap bound for the global operator. For weighted scalar-valued functions on facets, the ROL decomposition becomes the law of total variance, recovering classical spectral-gap bounds from local link expansion (Alev and Lau, STOC 2020). Moreover, for positive binary spin distributions, conditional influence describes the overlap of the corresponding subspaces, yielding known heat-bath gap bounds from spectral independence (Anari, Liu, and Oveis Gharan, FOCS 2020). The framework also offers a route toward quantum Gibbs sampling, with applications requiring model-specific geometric estimates and energy comparisons.
AI Summary
The paper asks how hierarchies of local energy problems can yield global spectral-gap bounds beyond scalar-valued functions. Recursive Operator Localization (ROL) defines effective quadratic energies by constrained minimization over full variables with prescribed linear readouts. When localization maps commute with readouts and full energies assemble exactly, the parent effective energy equals the weighted sum of child effective energies plus a nonnegative “gluing defect,” measuring the cost of reconciling independently minimizing local choices. Iteration gives an exact finite-depth decomposition. The quantitative result specializes to nested subspaces indexed by a weighted simplicial complex. Weighted projections onto removed directions define norm-loss parameters $\theta_\tau$. Child bounds $\theta_{\tau i}\leq c$ and an auxiliary classical link-walk gap $\lambda$ imply $\theta_\tau\leq (2-\lambda)c/[\lambda(1-c)]$, using a Hilbert-space spectral-criterion argument. Provided $c<\lambda/2$ at each recursive update, terminal coercivity bounds $\delta_\sigma$ yield a root gap bounded below by the minimum, over root-to-terminal paths, of $\delta_\sigma\prod_\tau(1-\overline\theta_\tau)$. The complete uniform specialization gives $\operatorname{gap}(L_\varnothing)\geq\delta_n[1-(n-1)\epsilon_n]/n$ under its overlap assumptions. Scalar facet functions recover total variance and known local-expansion gap bounds; binary conditional laws identify increment Gram matrices with normalized covariances, recovering the spectral-independence heat-bath gap product. Separate one-step estimates exploit the gluing defect quantitatively. Applications require verified subspace structure, overlap estimates, terminal bounds, and energy comparisons. General readouts additionally require full-space norm control. Quantum Gibbs sampling remains prospective, and recursive use of quantitative defect bounds is left open.
using GPT-5.6 Sol and GPT-6 Astra via Codex for brainstorming, developing and checking proofs, drafting and revising the exposition, preparing figures, and obtaining review-style feedback.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 56 pages