Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Vietoris–Rips stability (interleaving form).

Proved
BoltzmannBridge.vr_stability_interleaved

by raver1975 · Sep 11, 2026 · Mathlib c5ea003 (Lean v4.30.0)

aether-catalogapplications

Vietoris–Rips stability (interleaving form). If two distance matrices are uniformly within ε, their VR filtrations are ε-interleaved.

theorem BoltzmannBridge.vr_stability_interleaved(d₁ d₂ : α → α → ℝ) {ε : ℝ}
    (hε : 0 ≤ ε) (h : ∀ x y, |d₁ x y - d₂ x y| ≤ ε) :
    Filtration.Interleaved (diamFiltrationOf d₁) (diamFiltrationOf d₂) ε := by sorry

Formalization Note Transplanted verbatim from the Aether Catalog source Applications/BoltzmannBridge/BottleneckStability.lean; the statement is byte-identical to the source declaration, elaborated with autoImplicit disabled in the platform environment.

Preamble
-- Thm stub generated from Applications/BoltzmannBridge/BottleneckStability.lean
import Mathlib
import Definitions.Def_Applications_BoltzmannBridge_BottleneckStability
import Definitions.Def_Applications_BoltzmannBridge_HigherPersistence
/-
# The Boltzmann Bridge IV — The Interleaving Distance and Bottleneck Stability

This file closes the catalog's persistent-homology arc.  The earlier files built
the *filtration calculus* (`Applications.BoltzmannBridge.HigherPersistence`:
`Filtration`, `sublevelFaces`, `sublevel_mono`, the Vietoris–Rips `diamWeight`)
and the *relational interleaving lemmas*
(`Applications.BoltzmannBridge.PersistenceStability`: `stability_interleaving`,
`stability_compose`, `stability_two_sided`).  Those files produced a family of
scattered set-inclusion inequalities.  This file turns them into a single
coherent **metric theory of persistence stability**:

* a named, symmetric, additively-composable interleaving relation
  `Interleaved F G δ` (with `Interleaved_refl/symm/mono/trans`) — the relational
  skeleton of a graded preorder;
* a real-valued `interleavingDist`, shown to be a *symmetric, grounded
  pre-distance* (`interleavingDist_nonneg`, `interleavingDist_le`,
  `interleavingDist_self`, `interleavingDist_comm`);
* the Cohen-Steiner–Edelsbrunner–Harer sublevel stability theorem in sharp
  `1`-Lipschitz form: uniform `D`-closeness of the weights forces a
  `D`-interleaving and `interleavingDist ≤ D` (`stability_supDist`,
  `interleavingDist_le_supDist`);
* a Gromov–Hausdorff / correspondence-distortion layer over **explicit distance
  matrices** `d : α → α → ℝ` (`diamWeightOf`, `diamFiltrationOf`), resting on the
  single load-bearing estimate `diamWeightOf_dist_le` — *the simplex diameter is
  `1`-Lipschitz in the input metric* — yielding `vr_stability_interleaved` and
  `vr_stability_dist`;
* an end-to-end concrete certificate on two `3`-point clouds
  (`cloud_distortion`, `cloud_stability`, `cloud_interleavingDist_le`).

The entire stability phenomenon collapses onto one inequality: the simplex weight
is `1`-Lipschitz in the data.  Everything else is monotonicity bookkeeping.

## Main results

* `Interleaved_refl/symm/mono/trans` — interleaving is a graded preorder
* `interleavingDist_nonneg/le/self/comm` — a symmetric grounded pre-distance
* `stability_supDist`, `interleavingDist_le_supDist` — CESH `1`-Lipschitz stability
* `diamWeightOf_dist_le` — VR diameter is `1`-Lipschitz in the distance matrix
* `vr_stability_interleaved`, `vr_stability_dist` — distortion `≤ ε` ⇒ stability
* `cloud_distortion/stability/interleavingDist_le` — concrete point-cloud certificate
-/

open Finset BigOperators

open BoltzmannBridge

open Filtration

variable {α : Type*}

/-! ## The interleaving relation -/


-- !-- `0 ≤ 0`; and `F.sublevelFaces t ⊆ F.sublevelFaces (t+0)` simplifies via `t+0 = t`. -- !--

-- !-- Swap the two inclusion clauses; `0 ≤ δ` is preserved. -- !--

-- !-- Enlarge each shift via `sublevel_mono` (`t+δ ≤ t+δ'`); `0 ≤ δ ≤ δ'` by `linarith`. -- !--

-- !-- Chain the two interleavings' inclusions (cf. `stability_compose`); the shifts
-- !-- add since `t + (δ + δ') = (t + δ) + δ'`. -- !--

/-! ## The interleaving distance -/


-- !-- Every admissible shift is `≥ 0` (first component of `Interleaved`), so
-- !-- `Real.sInf_nonneg` gives the bound. -- !--

-- !-- `δ` lies in the shift set, which is bounded below by `0`; apply `csInf_le`. -- !--

-- !-- `≤ 0` from `interleavingDist_le` with `Interleaved_refl`, `≥ 0` from `nonneg`. -- !--

-- !-- `Interleaved_symm` makes the two shift sets equal, hence equal infima. -- !--

/-! ## Cohen-Steiner–Edelsbrunner–Harer sublevel stability (1-Lipschitz form) -/


-- !-- Each direction is `stability_two_sided`; the shift `D ≥ 0` packages the
-- !-- symmetric bound into an `Interleaved`. -- !--

-- !-- Combine `stability_supDist` with `interleavingDist_le`. -- !--


/-! ## Vietoris–Rips over explicit distance matrices -/


variable {α : Type*}


-- !-- `0` always sits in the inserted set, so the `sup'` dominates it via `le_sup'`. -- !--

-- !-- The empty product is empty, so the `sup'` is over `{0}`, giving `0`. -- !--

-- !-- Every pairwise distance of `σ` is a pairwise distance of `τ ⊇ σ`, so the
-- !-- smaller `sup'` is dominated by the larger one (`sup'_le` + `le_sup'`). -- !--


-- !-- The load-bearing `1`-Lipschitz estimate: bound each `d₁ x y ≤ d₂ x y + ε ≤`
-- !-- `diamWeightOf d₂ σ + ε`, and `0 ≤ diamWeightOf d₂ σ + ε`; then `sup'_le` both
-- !-- ways and `abs_sub_le_iff`. -- !--

-- !-- Apply `diamWeightOf_dist_le` to every `σ` to get `WeightCloseBy`, then
-- !-- `stability_supDist`. -- !--
Formal statement
theorem BoltzmannBridge.vr_stability_interleaved(d₁ d₂ : α → α → ℝ) {ε : ℝ}
    (hε : 0 ≤ ε) (h : ∀ x y, |d₁ x y - d₂ x y| ≤ ε) :
    Filtration.Interleaved (diamFiltrationOf d₁) (diamFiltrationOf d₂) ε := by sorry
Source
https://github.com/paulklemstine/Lean/blob/53c2925a02/Catalog/Applications/BoltzmannBridge/BottleneckStability.lean#L229

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, with reuse governed by our licensing terms.

How Prove2Me worksResearch paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me