Transfer uniform work-function growth bounds from finite subspaces
ProvedKServer.growth_bound_of_uniform_finite_subspacesFix , an arbitrary metric space , an initial configuration , and real numbers and . Assume there exists an injective -server configuration in .
Suppose that for every finite subset containing the initial configuration, and every request sequence in , the work function computed entirely in admits majorants of its injective increments whose sum is at most
The same constants must work for all such and .
Then every finite request sequence in admits majorants of all its injective work-function increments whose sum is at most
The target includes configurations using points outside the request sequence. No boundedness or compactness assumption on , existence of an optimal schedule, or attainment of maximal increments is imposed. The proof places finitely many selected terminal configurations and any feasible schedule into a common finite subspace, applies the finite bound, then takes an infimum over schedules and finite sums of suprema over terminal configurations. Constants depending on do not meet the hypothesis.
import Definitions.Def_KServer_workfunctionU open KServer
theorem KServer.growth_bound_of_uniform_finite_subspaces (k : ℕ) (hk : 1 ≤ k) (M : Type) [MetricSpace M]
(C₀ : Config k M) (a c : ℝ) (ha : 0 < a)
(hI : ∃ X : Config k M, Function.Injective X)
(H : ∀ (P : Finset M) (D₀ : Config k P),
(∀ i, (D₀ i : M) = C₀ i) → ∀ τ : List P, ∃ u : ℕ → ℝ,
(∀ t, t < τ.length → ∀ Y : Config k P, Function.Injective Y →
workFnU D₀ (τ.take (t + 1)) Y ≤ workFnU D₀ (τ.take t) Y + u t) ∧
(∑ t ∈ Finset.range τ.length, u t) ≤ a * offlineCost D₀ τ + c) :
∀ σ : List M, ∃ u : ℕ → ℝ,
(∀ t, t < σ.length → ∀ X : Config k M, Function.Injective X →
workFnU C₀ (σ.take (t + 1)) X ≤ workFnU C₀ (σ.take t) X + u t) ∧
(∑ t ∈ Finset.range σ.length, u t) ≤ a * offlineCost C₀ σ + c := by sorry