Proof of Theorem 5.2 — path expansion
ProvedROLDescent.path_expansionhilbert-space-operatorsp2o-gran-per-chapterp2o-plan-paperp2o-strengthenedp2o-v1simplicial-complexesspectral-gap
Let be a weighted pure complex with facet size . Let and be real functions on faces with at every internal face and
Then for every ,
the minimum being over root paths of steps (ordered sequences of legal additions), with the empty product equal to one. At this is the path formula (84) for .
Differences from the source:
- generalization: inside the proof of Theorem 5.2, under its hypotheses → only θ̄_τ < 1 at internal faces and (83), on a weighted pure complex; reason: the hypotheses of Theorem 5.2 give θ̄_τ < 1 (its first conclusion), so this specializes to the source's setting
- encoding: paths τ₀ = ∅ → τ₁ → ⋯ → τ_j; min over paths → rootPaths X j (injective vertex sequences with face prefixes), τ_k = prefixFace p k; ⨅ over this finite nonempty set; reason: a step τ → τi is determined by its added vertex; different orders remain different paths
Preamble
import Mathlib import Definitions.Def_ROLDescent_WeightedComplex import Definitions.Def_HDXConc_linkWeight import Definitions.Def_ROLDescent_prefixFace import Definitions.Def_ROLDescent_rootPaths
Formal statement
namespace ROLDescent
/-- **Path expansion of δ_∅.** Hayakawa, *Recursive Operator Localization and Spectral Gap Descent via the Geometry of Local
Spectral Expansion*, arXiv:2610.23695 (held version, sha256 702bd512…20bdd545), proof of Theorem 5.2, p.35 (PDF 35): "To make the path
formula explicit, write a root-to-stop path as p = (τ₀, …, τ_{d−1}), where τ₀ = ∅ and
τ_{d−1} = t(p). At each internal node, (83) reads δ_τ = min_{i∈N_τ} (1 − θ̄_τ)δ_τi. The retention
factor is positive, so substituting the same formula for each child lets us combine the
successive minima. More generally, after j steps, for 0 ≤ j ≤ d − 1,
δ_∅ = min_{τ₀=∅→τ₁→⋯→τ_j} { δ_τj ∏_{k=0}^{j−1} (1 − θ̄_τk) }, with the empty product equal to one."
**Formalization Note** A path of `j` steps is an element of `rootPaths X j` (an injective sequence of
added vertices with face prefixes); τ_k is `prefixFace p k`. The hypotheses are the weighted complex,
the positivity of the retention factors (θ̄_τ < 1 on T_int, `τ.card + 2 ≤ d`) and the recursion (83)
as an equation. The minimum over paths is an infimum over the finite, nonempty set of paths. -/
theorem path_expansion {V : Type*} [Fintype V] [DecidableEq V]
(X : Finset (Finset V)) (d : ℕ) (w : Finset V → ℝ) (hX : WeightedComplex X d w)
(θbar δ : Finset V → ℝ)
(hθ : ∀ τ ∈ X, τ.card + 2 ≤ d → θbar τ < 1)
(h83 : ∀ τ ∈ X, τ.card + 2 ≤ d →
δ τ = (1 - θbar τ) * ⨅ i : HDXConc.linkVerts X τ, δ (insert (i : V) τ)) :
∀ j ≤ d - 1, δ ∅ = ⨅ p : rootPaths X j,
δ (prefixFace p.1 j) * ∏ k ∈ Finset.range j, (1 - θbar (prefixFace p.1 k)) := by sorry
end ROLDescent
Source
Hayakawa, Recursive Operator Localization and Spectral Gap Descent via the Geometry of Local Spectral Expansion, arXiv:2610.23695 (held version, sha256 702bd512…20bdd545), proof of Theorem 5.2, p.35 (PDF 35)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.