orientation_maximin_value
def
The PROPm slack d_i: the maximum over other agents j of agent i's least value for a relevant edge held by j.
Statement
noncomputable def orientation_maximin_value {m n : ℕ} (ends : Fin m → Fin n × Fin n) (u : Fin n → Fin m → ℝ) (σ : Fin m → Fin n) (i : Fin n) : ℝ
- Defined in
- Fair Orientations: Proportionality and Equitability, arXiv:2602.18098 v2, Sec. 2 (relevant items, proportional share PROP_i = sum_{e in E_i} v_i(e)/n_e); PROPm as introduced at IJCAI 2021, with the maximin value restricted to relevant items as PROPX is in Def. 2.2; formalized by the operator (Claude), 2026-10-03
- Verified
- 03 Oct 2026
- Axioms
- none listed
- Built on 1
- relevant_min_valueAgent i's value for the least valuable of her relevant edges held by agent j, or 0 if j holds none.
- Used by 2
- is_propm_orientationAn orientation is PROPm if every agent's value for her bundle plus her PROPm slack d_i is at least her proportional share.
- propm_orientation_of_out_degree_twoIf the simple graph underlying a loopless multigraph has an orientation with every out-degree at most two, then for nonnegative additive…
Lean source view module on GitHub
/-- The PROPm slack `d_i = max_{j ≠ i} min { v_i(e) : e ∈ π_j ∩ E_i }` (the maximin value from the definition of PROPm, restricted to relevant items; `0` if no other agent holds a relevant item). -/ noncomputable def orientation_maximin_value {m n : ℕ} (ends : Fin m → Fin n × Fin n) (u : Fin n → Fin m → ℝ) (σ : Fin m → Fin n) (i : Fin n) : ℝ := (Finset.univ.erase i).fold max 0 fun j => relevant_min_value ends u σ i j