Lemma 5.1, pp. 12–13 — generating functions of ∆W, W, κ, π, ϖ: (1−s)/ψ, 1/ψ, φ/ψ, exp(±∫₀ˢ φ/ψ)
ProvedSkipFreeFluct.MBI.lemma_5_1branching-process-with-immigrationgenerating-functionp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1renewal-sequence
Let be MBI parameters, with mechanisms and . Let be the tails of and define recursively ,
, , , , .
Then , , and are nonnegative, and for the generating functions converge with
These sequences are the discrete analogues of scale functions; every identity of Theorem 5.4 is written in terms of them.
Differences from the source:
- encoding: parameter constraints of p. 12 (α > 0, µ, ν probability measures, µ(0) > 0, µ(1) = 0, p, β, q ≥ 0) → MBIParams.Valid, with HasSum µ 1, HasSum ν 1 and ν(0) = 0; reason: same constraints
- encoding: GF identities f*[s] = F(s) → HasSum (k ↦ f(k) s^k) (F(s)); reason: the generating function is the sum of this series
- encoding: '∆W : E → [0, ∞)', 'W, κ : E → [0, ∞)', 'the nonnegative measure π' → nonnegativity conclusion clauses; reason: the codomain claims of the lemma, stated as conclusions
Preamble
import Mathlib import Definitions.Def_SkipFreeFluct_MBI_Chain import Definitions.Def_SkipFreeFluct_MBI_Model open scoped ENNReal Topology open Filter MeasureTheory
Formal statement
namespace SkipFreeFluct.MBI
/-- Lemma 5.1, pp. 12–13. For an MBI with parameters `(α, µ, p, β, ν, q)`, the sequences `∆W`,
`W`, `κ`, `π` are nonnegative, and their generating functions and that of `ϖ` are, for
`s ∈ [0, s₀)`:
`∆W*[s] = (1 - s)/ψ(s)`, `W*[s] = 1/ψ(s)`, `κ*[s] = φ(s)/ψ(s)`,
`π*[s] = exp(∫_0^s φ(u)/ψ(u) du)`, `ϖ*[s] = exp(-∫_0^s φ(u)/ψ(u) du)`. -/
theorem lemma_5_1 (P : MBIParams) (hP : P.Valid) :
(∀ k : ℕ, 0 ≤ dW P k ∧ 0 ≤ W P k ∧ 0 ≤ kappa P k ∧ 0 ≤ piSeq P k) ∧
∀ s ∈ Set.Ico 0 (s0 P),
HasSum (fun k : ℕ => dW P k * s ^ k) ((1 - s) / psi P s) ∧
HasSum (fun k : ℕ => W P k * s ^ k) (1 / psi P s) ∧
HasSum (fun k : ℕ => kappa P k * s ^ k) (phiImm P s / psi P s) ∧
HasSum (fun k : ℕ => piSeq P k * s ^ k)
(Real.exp (∫ u in (0 : ℝ)..s, phiImm P u / psi P u)) ∧
HasSum (fun k : ℕ => varpi P k * s ^ k)
(Real.exp (-∫ u in (0 : ℝ)..s, phiImm P u / psi P u)) := by sorry
end SkipFreeFluct.MBI
Source
Loeffen, Patie, Wang, Fluctuation Theory of Continuous-Time, Skip-Free Downward Markov Chains with Applications to Branching Processes with Immigration, arXiv:2208.14425v1, pp. 12–13, Lemma 5.1
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.