Proof of Theorem 5, p. 215 — ḟ(t) = 2Ż̄·Z̄ = −2RṪ·Z̄ ≤ −2δ‖Z̄(t)‖ = −2δ√f(t)
ProvedMaxPressure.FluidStab.lyapunov_driftfluid-modellyapunov-functionp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1
Under the hypotheses of the previous statement (standing assumptions, a maximum pressure fluid model solution , and with ), at every regular time the Lyapunov function satisfies
This differential inequality is the core of the proof of Theorem 5: the energy decreases at a rate proportional to its square root, which forces it to reach zero in finite time.
Formalization Note is the Euclidean norm, written as the square root of the sum of squares. The statement gives the derivative of at and the chain of (in)equalities as conjuncts.
Preamble
import Mathlib import Definitions.Def_MaxPressure_FluidStab_Network
Formal statement
namespace MaxPressure.FluidStab
open Matrix
/-- Proof of Theorem 5, App. B, p. 215: under the hypotheses of `pressure_lower_bound`, at every
regular time `t` the Lyapunov function `f(t) = ∑_i Z̄_i(t)²` satisfies
`ḟ(t) = 2 Ż̄(t) · Z̄(t) = −2 R Ṫ(t) · Z̄(t) ≤ −2ε ‖Z̄(t)‖ = −2ε √f(t)`. -/
theorem lyapunov_drift {I J K : ℕ} (N : Network I J K) (hN : N.Standing)
(Zb : ℝ → Fin I → ℝ) (Tb : ℝ → Fin J → ℝ)
(hsol : IsFluidSolution N Zb Tb) (hmp : MPFluidEq N Zb Tb)
(xs : Fin J → ℝ) (hxs : xs ∈ allocSet N) (ε : ℝ) (hε : 0 < ε)
(hR : ∀ i, ε ≤ (R N *ᵥ xs) i) (t : ℝ) (ht : IsRegular Zb Tb t) :
HasDerivAt (fun s => ∑ i, Zb s i ^ 2) (2 * (deriv Zb t ⬝ᵥ Zb t)) t ∧
2 * (deriv Zb t ⬝ᵥ Zb t) = -2 * pressure N (deriv Tb t) (Zb t) ∧
-2 * pressure N (deriv Tb t) (Zb t) ≤ -2 * ε * Real.sqrt (∑ i, Zb t i ^ 2) := by sorry
end MaxPressure.FluidStab
Source
Dai & Lin, Maximum pressure policies in stochastic processing networks, Oper. Res. 53(2) (2005), p. 215, proof of Theorem 5 (App. B), second display
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.