Partial Lagrangian
DefinitionDSICCompComp_Supplement_lagrangianlp-dualitymechanism-designp2o-gran-per-chapterp2o-plan-paperp2o-v1revenue-maximization
Lagrangifying only the DSIC constraints of the revenue LP gives the partial Lagrangian
Formalization Note The bracket is dsicSlack, the same expression. Dual solutions are encoded as in FlowConservation.
Definition code
import Mathlib
import Definitions.Def_DSICCompComp_Supplement_jointPMF
import Definitions.Def_DSICCompComp_Supplement_dsicSlack
namespace DSICCompComp
namespace Supplement
/-- The partial Lagrangian `L(λ, x, p)` of the DSIC revenue LP, in which only the DSIC constraints are
Lagrangified (Ryu–Weinberg–Xue, *The DSIC Competition Complexity of Additive Buyers over Independent Items*, STOC 2027
submission, manuscript of Nov. 3, 2026, sha256 bf61bfdb…,
Appendix C, p.35 (pdf p.36)):
`L(λ, x, p) = ∑_{i, v ∈ supp(D)} f(v)·p_i(v) + ∑_{i, v ∈ supp(D), v'_i ∈ supp(D_i) ∪ {∅}} λ^{v_{-i}}_i(v_i, v'_i)·(∑_j v_ij·(x_ij(v) − x_ij(v'_i, v_{-i})) − (p_i(v) − p_i(v'_i, v_{-i})))`.
**Formalization Note** The bracket multiplying `λ^{v_{-i}}_i(v_i, v'_i)` is `dsicSlack` (the same
expression); dual solutions are encoded as in `FlowConservation`. -/
def lagrangian {n m : ℕ} (E : FiniteSetting n m)
(lam : (i : Fin n) → ((k : Fin n) → E.S k) → Option (E.S i) → ℝ) (sol : LPSolution E) : ℝ :=
∑ i, ∑ v : (k : Fin n) → E.S k, jointPMF E v * sol.p (fun k => some (v k)) i +
∑ i, ∑ v : (k : Fin n) → E.S k, ∑ w : Option (E.S i), lam i v w * dsicSlack E sol v i w
end Supplement
end DSICCompComp
Source
Ryu–Weinberg–Xue, The DSIC Competition Complexity of Additive Buyers over Independent Items, STOC 2027 submission (manuscript of Nov. 3, 2026, sha256 bf61bfdb…), Appendix C, p.35 (pdf p.36)