The forward guarantee (25):
DefinitionLawLocalComp_Supplement_ForwardContractlaw-compressionmonotone-normsp2o-gran-per-chapterp2o-plan-paperp2o-v1probabilitystochastic-loads
Let map finite nonnegative laws to finite nonnegative laws, let be a supplied resource (horizon , rank promise , or both) and . satisfies the forward guarantee with factor if for every dimension , every selection of independent jobs with rational laws and directions admitted by ( under a horizon, under a rank promise), and every norm monotone on , the loads and , with sampled independently per copy, satisfy
With this is (25). It is the contracting guarantee (1) with input and output exchanged.
Definition code
import Mathlib
import Definitions.Def_LawLocalComp_IsRatLaw
import Definitions.Def_LawLocalComp_IsMonotoneNorm
import Definitions.Def_LawLocalComp_expNorm
import Definitions.Def_LawLocalComp_dirRank
import Definitions.Def_LawLocalComp_Resource_admits
namespace LawLocalComp
namespace Supplement
/-- **The forward guarantee (25) with factor `κ`.**
Mahboub–Rahgoshay, *Optimal Law-Local Compression for Stochastic Loads*, STOC 2027
submission manuscript of Oct. 29, 2026 (sha256 5f71501c…aa873), Theorem 6.2, (25), p.10: `E g(L) ≤ E g(L_Ŷ) ≤ (1 + ε) E g(L)`.
`ForwardContract Φ R κ`: for every dimension `d`, every selection of `n` independent jobs with
rational laws `X_j` and nonnegative directions `a_j ∈ ℝ^d_+` admitted by the resource `R`, and every
norm `g` monotone on `ℝ^d_+`, the original load `L = ∑ X_j a_j` and the output load
`L_Ŷ = ∑ Φ(X_j) a_j` (outputs sampled independently per copy) satisfy
`E g(L) ≤ E g(L_Ŷ) ≤ κ E g(L)`. (25) is `κ = 1 + ε`.
**Formalization Note.** This is the published `Contract` with the roles of input and output
exchanged: here the output load is the larger one. -/
def ForwardContract (Φ : Law → Law) (R : Resource) (κ : ℝ) : Prop :=
∀ (d n : ℕ) (X : Fin n → Law) (a : Fin n → Fin d → ℝ), (∀ j, IsRatLaw (X j)) →
(∀ j, 0 ≤ a j) → R.admits n (dirRank a) → ∀ g : (Fin d → ℝ) → ℝ, IsMonotoneNorm g →
expNorm g X a ≤ expNorm g (fun j => Φ (X j)) a ∧
expNorm g (fun j => Φ (X j)) a ≤ κ * expNorm g X a
end Supplement
end LawLocalComp
Source
Mahboub–Rahgoshay, Optimal Law-Local Compression for Stochastic Loads, STOC 2027 submission (manuscript of Oct. 29, 2026, sha256 5f71501c…aa873), Theorem 6.2, (25), p.10