Reason: New lemma: the generalized mean-field flow is well defined on the L^2 control set and depends stably on its data. It gives bounds on the affine drift coefficient, independence of the flow from the choice of admissible representative, a quantitative estimate for the deviation of two flows in terms of the initial states and the weak pairing of the control difference against the drift coefficient, an equi-Lipschitz bound in time for that pairing, and uniform convergence of the flows when the initial states converge and the controls converge weakly.
Statement
Let (β0,β1) be an affine-controlled transition-rate family on l states with control set A⊆Rm, let β be its projected extension, and adopt from that lemma the real numbers R, K1, B and Λb, the aggregate state drift b, the projected drift b^, and the coefficient maps b0γ:Δl→R and b1γ:Δl→Rm of claim 3 there, where Δl is the probability simplex. Let T>0 be a real number and write Kb=2l(l−1)B.
1. (Bounds on the affine coefficient.) For every γ∈{1,…,l} and every Σ∈Δl we have ∣b1γ(Σ)∣≤C1, and
b1γ(Σ)⋅(α′−α)≤K2∣α′−α∣≤L1for all α,α′∈A.
2. (Admissible representatives and the flow.) Every ξ∈UA has a representative u with u(t)∈A for everyt∈[0,T]; such a u is called an admissible representative of ξ. Let x0∈Δl and ξ∈UA. For an admissible representative u of ξ, let Su be the map furnished by claim 1 of the existence and uniqueness theorem applied to the initial value x0 and the control u. Then Su does not depend on the choice of admissible representative u of ξ; write S(x0,ξ) for it, with values St(x0,ξ). It satisfies St(x0,ξ)∈Δl for every t, S0(x0,ξ)=x0, and ∣St(x0,ξ)−Sr(x0,ξ)∣≤Kb∣t−r∣ for all r,t∈[0,T].
3. (The control-perturbation functionals.) Let x0∈Δl and ξ∈UA, and write S=S(x0,ξ). For γ∈{1,…,l} and t∈[0,T], the map [0,T]→Rm sending s to 1[0,t](s)b1γ(Ss) is square-integrable; let wγ,t∈H denote its class. For ζ∈UA put
gtγ(ζ)=⟨ζ−ξ,wγ,t⟩L2.
4. (Quantitative stability.) In the setting of claim 3, let x0′∈Δl and ξ′∈UA and write S′=S(x0′,ξ′). Let G be a real number with ∣grγ(ξ′)∣≤G for every γ∈{1,…,l} and every r∈[0,T]. Then
∣St′−St∣≤eΛbT(∣x0′−x0∣+lG)for every t∈[0,T],
5. (Equicontinuity in time.) In the setting of claim 3, for every ζ∈UA, every γ∈{1,…,l} and all r,t∈[0,T],
gtγ(ζ)−grγ(ζ)≤L1∣t−r∣.
The constant L1 depends only on the rate family and A, not on ζ, on ξ or on x0.
6. (Sequential stability.) Let (x0j)j∈N be a sequence in Δl and let x0∈Δl be such that the real sequence (∣x0j−x0∣)j∈N has limit0. Let (ξj)j∈N be a sequence in UA and let ξ∈UA be such that ξj⇀ξ in the sense of weak convergence. Then, writing Sj=S(x0j,ξj) and S=S(x0,ξ), for every real ε>0 there is N∈N such that
Curated associations between results. These are editable and subjective — they do not replace the dependency graph, which is derived from the references in the text.