Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Lemma 3.2 — the maximum of linear Lyapunov components has drift ≤ −min εᵢ

Proved
DaiWeissFluid.ThreeBuffer.lemma3_2

by mikedeng1 · Oct 5, 2026 · Mathlib 0df444a (Lean v4.33.1)

fluid-modellyapunov-methodsp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1reentrant-line

Consider a reentrant line with I≥1I \ge 1I≥1 stations and KKK classes, mean service times mk>0m_k > 0mk​>0, and a fluid model solution (Q,T)(Q, T)(Q,T) satisfying (1.8)–(1.12). For each station iii let Gi(t)=∑kcikQk(t)G_i(t) = \sum_{k} c_{ik} Q_k(t)Gi​(t)=∑k​cik​Qk​(t) be a linear function of the fluid levels with nonnegative coefficients cik≥0c_{ik} \ge 0cik​≥0, and let Wi(t)=∑k∈CimkQk(t)W_i(t) = \sum_{k \in C_i} m_k Q_k(t)Wi​(t)=∑k∈Ci​​mk​Qk​(t) be the immediate volume of station iii. Assume:

  1. (a) for each iii there is εi>0\varepsilon_i > 0εi​>0 such that, at every t>0t > 0t>0 with Wi(t)>0W_i(t) > 0Wi​(t)>0 at which GiG_iGi​ is differentiable, G˙i(t)≤−εi\dot G_i(t) \le -\varepsilon_iG˙i​(t)≤−εi​;
  2. (b) for each iii and each t≥0t \ge 0t≥0, if Wi(t)=0W_i(t) = 0Wi​(t)=0 then Gi(t)≤Gj(t)G_i(t) \le G_j(t)Gi​(t)≤Gj​(t) for every j≠ij \ne ij=i.

Let G(t)=max⁡{G1(t),…,GI(t)}G(t) = \max\{G_1(t), \dots, G_I(t)\}G(t)=max{G1​(t),…,GI​(t)}. Then GGG is absolutely continuous on every compact interval of [0,∞)[0,\infty)[0,∞) and nonnegative there, and at every t>0t > 0t>0 at which G,G1,…,GIG, G_1, \dots, G_IG,G1​,…,GI​ are all differentiable and G(t)>0G(t) > 0G(t)>0,

G˙(t)≤−ε,ε=min⁡{ε1,…,εI}.\dot G(t) \le -\varepsilon, \qquad \varepsilon = \min\{\varepsilon_1, \dots, \varepsilon_I\}.G˙(t)≤−ε,ε=min{ε1​,…,εI​}.

The lemma turns station-wise drift conditions into a single drift bound for a piecewise-linear Lyapunov function; combined with Lemma 2.2 it yields stability of a fluid model.

Formalization Note. Stations and classes are 0-based. "Nonnegative linear function of Q(t)Q(t)Q(t)" is encoded by time-independent nonnegative coefficients c i k. The maximum is ⨆ i : Fin I, a supremum over a finite nonempty index type (the hypothesis 0<I0 < I0<I makes it a genuine maximum and makes the minimum of the εi\varepsilon_iεi​ well defined, written Finset.univ.inf' _ ε). The lemma assumes only (1.8)–(1.12) and mk>0m_k > 0mk​>0, not work conservation, as in the paper. A "regular point" is a time t>0t>0t>0 at which the relevant functions are differentiable (HasDerivAt, DifferentiableAt).

Preamble
import Mathlib
import Definitions.Def_DaiWeissFluid_ThreeBuffer_FluidModel
Formal statement
namespace DaiWeissFluid.ThreeBuffer

/-- Lemma 3.2, p. 121, for an arbitrary reentrant line with `I ≥ 1` stations: let
`G_i(t) = ∑_k c i k * Q_k(t)` with nonnegative coefficients and `G(t) = max_i G_i(t)`. If
(a) `W_i(t) > 0` implies `Ġ_i(t) ≤ -ε_i`, and (b) `W_i(t) = 0` implies `G_i(t) ≤ G_j(t)` for all
`j ≠ i`, then `G` is absolutely continuous and nonnegative on `[0, ∞)`, and at every regular point
`t > 0` of `G, G_1, …, G_I` with `G(t) > 0`, `Ġ(t) ≤ -min_i ε_i`. -/
theorem lemma3_2 {I K : ℕ} (L : ReentrantLine I K) (hm : ∀ k, 0 < L.m k) (hI : 0 < I)
    (Q T : ℝ → Fin K → ℝ) (hsol : L.IsFluidSolution Q T)
    (c : Fin I → Fin K → ℝ) (hc : ∀ i k, 0 ≤ c i k) (ε : Fin I → ℝ) (hε : ∀ i, 0 < ε i)
    (ha : ∀ i t, 0 < t → 0 < L.volume Q i t →
      ∀ d, HasDerivAt (fun s => ∑ k, c i k * Q s k) d t → d ≤ -ε i)
    (hb : ∀ i t, 0 ≤ t → L.volume Q i t = 0 →
      ∀ j, j ≠ i → ∑ k, c i k * Q t k ≤ ∑ k, c j k * Q t k) :
    (∀ a b, 0 ≤ a → a ≤ b →
        AbsolutelyContinuousOnInterval (fun s => ⨆ i, ∑ k, c i k * Q s k) a b) ∧
    (∀ t, 0 ≤ t → 0 ≤ ⨆ i, ∑ k, c i k * Q t k) ∧
    (∀ t, 0 < t → 0 < (⨆ i, ∑ k, c i k * Q t k) →
      (∀ i, DifferentiableAt ℝ (fun s => ∑ k, c i k * Q s k) t) →
      ∀ d, HasDerivAt (fun s => ⨆ i, ∑ k, c i k * Q s k) d t →
        d ≤ -(Finset.univ.inf' (Finset.univ_nonempty_iff.mpr ⟨⟨0, hI⟩⟩) ε)) := by sorry

end DaiWeissFluid.ThreeBuffer
Source
Dai and Weiss, Stability and instability of fluid models for reentrant lines, Math. Oper. Res. 21(1) (1996), pp. 121–122, Lemma 3.2
Human review
  • Endorsed by Shuze Chen · Oct 6, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 6, 2026

    Confirmed by the mission captain (proposal self-audit).

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