Power cost for real
DefinitionKMeansTaper_Supplement_powerCostapproximation-algorithmsfacility-locationk-meanslinear-programmingp2o-gran-per-chapterp2o-plan-paperp2o-v1
Let be a finite rational metric instance and let be a real exponent. For a nonempty set of candidate facilities, the power cost is
This is the general objective introduced before Theorem 3.6. At it is exactly the published squared objective KMeansTaper.cost2.
Formalization Note Real exponentiation is Mathlib's Real.rpow. The reused nearest returns zero on the empty set, where the paper's distance is infinite; theorem statements therefore require to be nonempty.
Definition code
import Mathlib
import Definitions.Def_MetricKMedian
import Definitions.Def_KMeansTaper_cost2
namespace KMeansTaper
namespace Supplement
open OAI.MetricKMedian (Instance nearest)
/-- **Power cost** `cost_p(S) = ∑_{j ∈ J} d(j,S)^p`, introduced immediately before
Theorem 3.6, p.5 (PDF p.8).
Zhu, *A tight approximation for metric k-means via squared tapered rounding*, STOC 2027
submission (manuscript sha256 8b6610229e43dc2441325b120437136afe762dbd6e873f73919fcb6b043fc5a3).
**Formalization Note.** The exponent is a real number and exponentiation is `Real.rpow`.
`nearest I j S` is `0` when `S` is empty, whereas the paper's distance to the empty set is
infinite; every theorem using `powerCost` therefore requires `S.Nonempty`. -/
noncomputable def powerCost (I : Instance) (p : ℝ) (S : Finset (Fin I.n)) : ℝ :=
∑ j ∈ I.clients, (nearest I j S) ^ p
/-- At exponent two, `powerCost` is exactly the published squared cost `cost2`. -/
@[simp] theorem powerCost_two (I : Instance) (S : Finset (Fin I.n)) :
powerCost I 2 S = cost2 I S := by
simp [powerCost, cost2]
end Supplement
end KMeansTaper
Source
Zhu, A tight approximation for metric k-means via squared tapered rounding, STOC 2027 submission (manuscript sha256 8b6610229e43dc2441325b120437136afe762dbd6e873f73919fcb6b043fc5a3), paragraph before Theorem 3.6, p.5 (PDF p.8)