While this submission is a draft, it cannot be used by other submissions.

High edge cost forces a strict entropy decrease

Lax263307.EntropyDecrease · concepts/Lax263307/EntropyDecrease.lean · lax-263307

proven

Loading review…

Sign in with ORCID

Community review

Flags

Each flag is tied to a public ORCID identity and explains why this concept may be incorrect.

No flags have been submitted.

    Community review

    Flag this concept

    State precisely what appears incorrect. This explanation will be public under your ORCID name.

    No source line selected.

    Natural Language Statement

    Theorem

    Let y∈Rr(G,k)y\in R_r(G,k), k>0k>0, ∣A∣<r|A|<r, and yA>0y_A>0. For a positive integer pp, if the conditioned costs satisfy

    ∑uv∈E(G)xuxv>k2/p,\sum_{uv\in E(G)}x_u x_v>k^2/p,

    then some vertex vv has πv>0\pi_v>0 and conditioning on vv decreases entropy by more than 2/p2/p:

    H(πv)<H(π)−2/p.H(\pi^v)<H(\pi)-2/p.
    Concept map
    6 concepts
    100%
    Proven claimDefinitionThis conceptRelated conceptA → B: B builds on ADescendants are omitted for concepts with more than 10 descendants.
    Evidence

    Each proof establishes this claim relative to its assumptions.

    Lean source view on GitHub

    1import Lax263307.ConditioningDistribution
    2import Lax263307.EntropyGap
    3
    4/-!
    5---
    6title: High edge cost forces a strict entropy decrease
    7type: theorem
    8---
    9Let y∈Rr(G,k)y\in R_r(G,k), k>0k>0, ∣A∣<r|A|<r, and yA>0y_A>0.
    10For a positive integer pp, if the conditioned costs satisfy
    11∑uv∈E(G)xuxv>k2/p,\sum_{uv\in E(G)}x_u x_v>k^2/p,
    12then some vertex vv has πv>0\pi_v>0 and conditioning on vv decreases entropy
    13by more than 2/p2/p:
    14H(πv)<H(π)−2/p.H(\pi^v)<H(\pi)-2/p.
    15
    16# Attribution
    17
    18Bonnet–Chang, proof of Lemma 12, the consequence of equation (18).
    19-/
    20
    21namespace Lax263307.EntropyDecrease
    22
    23open LinearRelaxation ConditioningDistribution Entropy
    24
    25/-- An expensive edge product forces a positive-cost conditioning step. -/
    26axiom exists_entropy_decrease {n : ℕ} (G : SimpleGraph (Fin n)) [DecidableRel G.Adj]
    27 (k r p : ℕ) (y : MomentVector n r) (feasible : IsFeasible G k r y)
    28 (A : Finset (Fin n)) (size : A.card < r) (positive : 0 < value y A)
    29 (target_positive : 0 < k) (threshold_positive : 0 < p)
    30 (expensive : (k : ℝ) ^ 2 / p < edgeCost G (conditioned y A)) :
    31 ∃ v : Fin n, 0 < marginal y A k v ∧
    32 entropy (marginal y (insert v A) k) < entropy (marginal y A k) - 2 / p
    33
    34end Lax263307.EntropyDecrease
    35
    Show Proof

    Attribution

    Bonnet–Chang, proof of Lemma 12, the consequence of equation (18).

    Discussion

    Ask a question or add context. Endorsements and structured flags are kept in the review panel above.

    Loading discussion…