Lemma 12 (used in the proof of Theorem 4) — ten-task potential base
OpenRealPinwheel.FiveSixths.Supplement.potential_basep2o-batch-stoc27sp2o-gran-per-chapterp2o-plan-paperp2o-v1pinwheel-schedulingreal-periods
Let be a nonempty ordered real-period instance with . Periods may be irrational, and equal periods label distinct tasks. Write for density and for potential. Schedulability means that a single schedule assigns exactly one task to each day and, for every task , positive integer and starting day , executes at least times in .
In dimension , if and , then
This is the full ten-task potential base case supporting Theorem 4. Potential omits only the final task's density; the hypothesis is not a bound on total density. The conclusion covers both branches of Appendix A and their equality boundaries, without any certificate or branch hypothesis.
Differences from the source:
- encoding: positive days and task labels 1 through n → natural days starting at zero and task positions Fin n; reason: day t encodes source day t+1 and position i encodes source task i+1; all windows and distinct task identities are preserved
Preamble
import Definitions.Def_RealPinwheel_FiveSixths_Instance
Formal statement
namespace RealPinwheel.FiveSixths.Supplement
/-- Lemma 12, printed p. 11 (PDF p. 12); repeated and proved in Appendix A,
printed p. 16 (PDF p. 17). Covers the entire ten-task potential domain. -/
theorem potential_base (A : Instance 10)
(hmin : 4 ≤ A.period ⟨0, A.positive_dimension⟩)
(hP : potential A ≤ (17 : ℝ) / 20) :
Schedulable A.period := by sorry
end RealPinwheel.FiveSixths.Supplement
Source
Hiroshi Fujiwara, Grzegorz Gutowski, Katsuhisa Ouchi, Alexander Wolff, STOC 2027 submission #885 (October 2026), Lemma 12, printed p. 11 (PDF p. 12); proof printed p. 16 (PDF p. 17)