graph_lly_curvature_tendsto
In a finite connected graph, for x ≠ y, κ_α(x, y)/(1 − α) converges as α → 1⁻, to κ_LLY(x, y): the limit in the definition of the Lin–Lu–Yau curvature exists.
Statement
theorem graph_lly_curvature_tendsto {V : Type*} [Fintype V] [DecidableEq V] (G : SimpleGraph V) [DecidableRel G.Adj] (hG : G.Connected) (x y : V) (hxy : x ≠ y) : Filter.Tendsto (fun α => graph_ollivier_curvature G α x y / (1 - α)) (nhdsWithin 1 (Set.Iio 1)) (nhds (graph_lly_curvature G x y))
- Formalizes
- arXiv:2610.10559 (Edge-Connectivity versus Lin--Lu--Yau Curvature), Sec. 2.2: existence of the limit in Definition 2.3 (quoted there from its reference [15]); finite connected graphs.
- Verified
- 10 Oct 2026
- Axioms
Classical.choiceQuot.soundpropext- Built on 4
- graph_ollivier_curvature_div_monotoneOnIn a finite connected graph, for x ≠ y, α ↦ κ_α(x, y)/(1 − α) is nondecreasing on (0, 1).
- graph_ollivier_curvature_div_le_two_div_distIn a finite connected graph, for x ≠ y and α ∈ [0, 1), κ_α(x, y)/(1 − α) ≤ 2/d(x, y).
- graph_lly_curvatureThe Lin–Lu–Yau curvature κ_LLY(x, y) = lim_{α → 1⁻} κ_α(x, y)/(1 − α).
- graph_ollivier_curvatureThe α-Ollivier–Ricci curvature κ_α(x, y) = 1 − W(μ_x^α, μ_y^α)/d(x, y).
- Used by 3
- graph_lly_curvature_le_two_div_distFor distinct vertices x, y of a finite connected graph, κ_LLY(x, y) ≤ 2/d(x, y).
- graph_lly_curvature_le_of_lipschitz_cut_potentialKantorovich-duality bound on the Lin–Lu–Yau curvature at a pair: for x ≠ y and any 1-Lipschitz potential f with f(x) − f(y) = d(x,y),…
- graph_lly_curvature_pos_of_wasserstein_lt_oneIf the Wasserstein distance between the lazy random-walk measures μ_x^α and μ_y^α on an edge xy is strictly less than 1 for some α ∈ (0,…
The proof, explained
The Lin–Lu–Yau curvature is defined as the limit of the rescaled Ollivier curvature κ_α(x, y) / (1 - α) as α → 1⁻. The rescaled curvature is monotone increasing on (0, 1) and bounded from above by 2 / d(x, y), so by the monotone convergence theorem for left limits (`MonotoneOn.tendsto_nhdsWithin_Ioo_left`), it converges as α → 1⁻ to its supremum. Unfolding the definition of `graph_lly_curvature` and applying `h.limUnder_eq` equates the definition with this limit.
The rescaled curvature α ↦ κ_α(x, y) / (1 - α) converges as α → 1⁻ to the supremum of its values on (0, 1).
We apply the monotone convergence theorem for left limits on an open interval (`MonotoneOn.tendsto_nhdsWithin_Ioo_left` with lower endpoint 0 and upper endpoint 1). The interval (0, 1) is non-empty because it contains 1/2 (checked by `norm_num`). Monotonicity of the quotient on (0, 1) is provided by `graph_ollivier_curvature_div_monotoneOn`. Finally, 2 / d(x, y) is an upper bound on (0, 1) by `graph_ollivier_curvature_div_le_two_div_dist`.
uses
MonotoneOn.tendsto_nhdsWithin_Ioo_left,graph_ollivier_curvature_div_monotoneOn,graph_ollivier_curvature_div_le_two_div_disthave h := MonotoneOn.tendsto_nhdsWithin_Ioo_left (y := (0 : ℝ)) (x := 1) ⟨1 / 2, by norm_num, by norm_num⟩ (graph_ollivier_curvature_div_monotoneOn G hG x y hxy) ⟨2 / (G.dist x y : ℝ), by rintro _ ⟨α, hα, rfl⟩ exact graph_ollivier_curvature_div_le_two_div_dist G hG x y hxy α hα.1.le hα.2⟩
The rescaled curvature converges to the Lin–Lu–Yau curvature κ_LLY(x, y) as α → 1⁻.
We unfold the definition of `graph_lly_curvature`, which is defined as the filter limit (`limUnder`) of the rescaled curvature as α → 1⁻. Since convergence to the supremum was established in the hypothesis h, `h.limUnder_eq` identifies this `limUnder` with that supremum limit, which closes the goal with h.
uses
h.limUnder_equnfold graph_lly_curvature rw [h.limUnder_eq] exact h
MonotoneOn.tendsto_nhdsWithin_Ioo_left- A function that is monotone on a non-empty open interval (y, x) and bounded above on that interval converges as its argument approaches x from the left to the supremum of its values on (y, x).
graph_ollivier_curvature_div_monotoneOn- In a connected graph, for any two distinct vertices x and y, the rescaled Ollivier curvature α ↦ κ_α(x, y)/(1 - α) is monotone increasing on the interval (0, 1).
graph_ollivier_curvature_div_le_two_div_dist- In a connected graph, for any two distinct vertices x and y and any α ∈ [0, 1), the rescaled Ollivier curvature κ_α(x, y)/(1 - α) is bounded above by 2/d(x, y).
h.limUnder_eq- If a function converges to a point along a filter in a Hausdorff space, its `limUnder` along that filter equals that limit point.
- In Lean, `graph_lly_curvature` is defined using `Filter.limUnder`, which formally assigns a value even if the limit were not to exist. This theorem proves that the limit does exist, and `h.limUnder_eq` connects the formal `limUnder` expression to the topological limit.
Written by a model (gemini-3.8-flash) from the Lean proof and the goal states Lean reported at each step. Lean verified the proof; this explanation is a reading aid and is not itself machine-checked.
Lean source view module on GitHub
Choose a tactic — a marked line — to see the goals Lean had before and after it, as in Lean’s infoview. Step with ← → or j k.
theorem graph_lly_curvature_tendsto {V : Type*} [Fintype V] [DecidableEq V] (G : SimpleGraph V) [DecidableRel G.Adj] (hG : G.Connected) (x y : V) (hxy : x ≠ y) : Filter.Tendsto (fun α => graph_ollivier_curvature G α x y / (1 - α)) (nhdsWithin 1 (Set.Iio 1)) (nhds (graph_lly_curvature G x y)) := by have h := MonotoneOn.tendsto_nhdsWithin_Ioo_left (y := (0 : ℝ)) (x := 1) ⟨1 / 2, by norm_num, by norm_num⟩ (graph_ollivier_curvature_div_monotoneOn G hG x y hxy) ⟨2 / (G.dist x y : ℝ), by rintro _ ⟨α, hα, rfl⟩ exact graph_ollivier_curvature_div_le_two_div_dist G hG x y hxy α hα.1.le hα.2⟩ unfold graph_lly_curvature rw [h.limUnder_eq] exact h