Excess running supremum after a peak equals the reflected shifted-path maximum
ProvedAvramDividend.Classical.runningSup_excess_after_peakavram-classicalbarrier-dividendsreflectionrunning-supremumtime-shift
For any bounded-above real-valued path f on nonnegative times, if its running supremum at time u is attained at f(u), then the further increase in running supremum by time u+t equals the positive part of the shifted interval supremum minus f(u). This is the deterministic incremental-dividend identity at an attained barrier crossing before applying strong Markov.
Preamble
import Mathlib open scoped NNReal ENNReal
Formal statement
theorem AvramDividend.Classical.runningSup_excess_after_peak
(f : ℝ≥0 → ℝ) (u t : ℝ≥0)
(hb : BddAbove (f '' Set.Icc 0 (u + t)))
(hpeak : sSup (f '' Set.Icc 0 u) = f u) :
sSup (f '' Set.Icc 0 (u + t)) - f u =
max 0 (sSup ((fun r : ℝ≥0 => f (u + r)) '' Set.Icc 0 t) - f u) := by sorrySource
Pathwise consequences of the separately published interval splitting runningSup_split_at and time translation runningSup_interval_shift lemmas, plus lattice maximum algebra. No stopping time or law equivalence assumed.