AFTD Auto-Formalizing Theoretical Domains

dephased_tree_local_noise_criterion

def

For a finite simple graph T on vertex type V and noise profile q : V → ℝ, dephased_tree_local_noise_criterion T q holds if both local conditions hold: (i) at every degree-two vertex v with distinct neighbors u and w, q v ≥ q u * q w; (ii) at every vertex v of degree at least three, at most one neighbor u has q u > 0 (that is, any two neighbors u and w with q u > 0 and q w > 0 must be equal).

Statement

def dephased_tree_local_noise_criterion {V : Type*} [Fintype V] [DecidableEq V]
    (T : SimpleGraph V) [DecidableRel T.Adj] (q : V → ℝ) : Prop
Defined in
Fuentes et al., arXiv:2511.19585; OP-827
Verified
02 Nov 2026
Axioms
none listed
Used by 3

Read back from the Lean

Given a finite type V with decidable equality, a simple graph T on V with decidable adjacency, and a function q : V → ℝ, dephased_tree_local_noise_criterion T q asserts that both of the following conditions hold: 1. For every vertex v : V with T.degree v = 2, and for all vertices u, w : V such that u ≠ w, T.Adj v u, and T.Adj v w, we have q v ≥ q u * q w. 2. For every vertex v : V with T.degree v ≥ 3, and for all vertices u, w : V such that T.Adj v u, T.Adj v w, 0 < q u, and 0 < q w, we have u = w (that is, v has at most one neighbor with strictly positive value under q).

Written by a model that saw only the Lean, never the English above. If the two disagree, that is worth a challenge.

Lean source view module on GitHub

/-- Local noise criterion for MMI on an inhomogeneously dephased tree. -/
def dephased_tree_local_noise_criterion {V : Type*} [Fintype V] [DecidableEq V]
    (T : SimpleGraph V) [DecidableRel T.Adj] (q : V → ℝ) : Prop := (∀ (v : V), T.degree v = 2 →
    ∀ (u w : V), u ≠ w → T.Adj v u → T.Adj v w → q v ≥ q u * q w) ∧
  (∀ (v : V), 3 ≤ T.degree v →
    ∀ (u w : V), T.Adj v u → T.Adj v w → 0 < q u → 0 < q w → u = w)
See it in the graphDiscuss this resultChallenge it