(3.6) — for lsc proper convex on a Banach space, implies
ProvedRockafellarMaxMono.Cyclic.subdiff_subset_eq_add_constLet be a real Banach space with dual , and let and be lower semicontinuous proper convex functions on such that
Then there is a real constant with
Rockafellar notes that, given Theorem 1 of his 1966 paper and its Corollary 2, this statement is exactly what remains to prove Theorem B. It yields both the maximality of among cyclically monotone operators and the uniqueness of up to an additive constant. The hypothesis is a one-sided inclusion.
Formalization Note "" is non-strict inclusion. The constant is a real number (not ); at points where the equation reads .
import Mathlib import Definitions.Def_RockafellarMaxMono_Shared_ProperConvex import Definitions.Def_RockafellarMaxMono_Shared_Subdiff
namespace RockafellarMaxMono.Cyclic
theorem subdiff_subset_eq_add_const {E : Type*} [NormedAddCommGroup E] [NormedSpace ℝ E]
[CompleteSpace E] (f g : E → EReal)
(hf : Shared.ProperConvex f) (hflsc : LowerSemicontinuous f)
(hg : Shared.ProperConvex g) (hglsc : LowerSemicontinuous g)
(hsub : ∀ x : E, Shared.subdiff f x ⊆ Shared.subdiff g x) :
∃ c : ℝ, ∀ x : E, g x = f x + (c : EReal) := by sorry
end RockafellarMaxMono.Cyclic
Read-back
What the Lean code literally says, in plain math · claude-opus-5-5
Setting.
- is a real Banach space. is its continuous dual.
- each satisfy the bundle's "proper convex" predicate (code not supplied).
- and are each lower semicontinuous in the norm topology.
- is the bundle's subdifferential operation, with values that are subsets of . Its code is not supplied.
Hypothesis. For every :
Conclusion. There is a real (finite) constant such that, for every and as an equality of extended reals:
Degenerate cases.
- Adding a finite leaves unchanged. So wherever the conclusion forces , and wherever it forces . In particular, and are asserted to take infinite values at exactly the same points.
- If , the conclusion says that is a finite real number when both values are finite, or that when either is infinite.
- Whether the inclusion hypothesis can hold vacuously depends on the unseen subdifferential definition. For example, it would if were empty everywhere.
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.