An average lost-sales bound controls ordinary base-stock cost
ProvedCappedBaseStock.base_stock_cost_of_loss_boundConsider the zero-start lost-sales inventory model with i.i.d. nonnegative demand of finite positive mean , lead time , and holding and lost-sales rates . Fix any ordinary base-stock level . Let denote lost sales, and let be the upper limit of expected average period costs from the empty initial state.
For every finite real satisfying
the cost satisfies
This transfers any bound on average lost sales to a total-cost bound at the same stock level. Both averages use the actual zero-start trajectory; no stationary convergence hypothesis is included. The positive-part notation is the exact embedding of the real expression into the model's extended nonnegative cost space.
Formalization Note The real affine expression is formed before applying ENNReal.ofReal. Its intercept can be negative and is not separately truncated.
import Definitions.Def_CappedBaseStock_BaseStockAnalysis open scoped ENNReal NNReal
namespace CappedBaseStock
theorem base_stock_cost_of_loss_bound (P : DemandLaw) (c : Parameters)
(S : ℝ≥0) (ell : ℝ) (hell : 0 ≤ ell)
(hloss : baseStockAverageLoss P c S ≤ ENNReal.ofReal ell) :
baseStockCost P c S ≤ ENNReal.ofReal
(c.h * ((S : ℝ) - ((c.L : ℝ) + 1) * mean P) +
(c.p + c.h * ((c.L : ℝ) + 1)) * ell) := by sorry
end CappedBaseStock