Separation Hahn-Banach theorem #
In this file we prove the geometric Hahn-Banach theorem. For any two disjoint convex sets, there exists a continuous linear functional separating them, geometrically meaning that we can intercalate a plane between them.
We provide many variations to stricten the result under more assumptions on the convex sets:
geometric_hahn_banach_open: One set is open. Weak separation.geometric_hahn_banach_open_point,geometric_hahn_banach_point_open: One set is open, the other is a singleton. Weak separation.geometric_hahn_banach_open_open: Both sets are open. Semistrict separation.geometric_hahn_banach_of_nonempty_interior': One set has nonempty interior. Nonstrict separation.geometric_hahn_banach_of_nonempty_interior: One set has nonempty interior, the other one is nonempty. Nonstrict separation by a nonzero functional.geometric_hahn_banach_of_nonempty_interior_point: One set has nonempty interior, the other one is a singleton outside this interior. Nonstrict separation, with the maximum attained at the singleton.geometric_hahn_banach_compact_closed,geometric_hahn_banach_closed_compact: One set is closed, the other one is compact. Strict separation.geometric_hahn_banach_point_closed,geometric_hahn_banach_closed_point: One set is closed, the other one is a singleton. Strict separation.geometric_hahn_banach_point_point: Both sets are singletons. Strict separation.
Given a set s which is a convex neighbourhood of 0 and a point x₀ outside of it, there is
a continuous linear functional f separating x₀ and s, in the sense that it sends x₀ to 1 and
all of s to values strictly below 1.
A version of the Hahn-Banach theorem: given disjoint convex sets s, t where s is open,
there is a continuous linear functional which separates them.
If s and t are convex, interior s is nonempty and disjoint from t, then a nonzero
continuous linear functional weakly separates s and t. The proof first separates interior s
from t, then extends the inequality from interior s to all of s using
closure (interior s) = closure s.
If s and t are convex, interior s is nonempty and disjoint from t, then a continuous
linear functional weakly separates s and t. If t is nonempty, this follows from
geometric_hahn_banach_of_nonempty_interior; if t = ∅, the zero functional works.
If A is convex with nonempty interior and x ∉ interior A, then there is a nonzero
continuous linear functional whose maximum on A is attained at x.
A version of the Hahn-Banach theorem: given disjoint convex sets s, t where s is
compact and t is closed, there is a continuous linear functional which strongly separates them.
A version of the Hahn-Banach theorem: given disjoint convex sets s, t where s is
closed, and t is compact, there is a continuous linear functional which strongly separates them.
See also NormedSpace.eq_iff_forall_dual_eq.
A closed convex set is the intersection of the half-spaces containing it.
If s and t are convex, interior s is nonempty and disjoint from t, then a continuous
𝕜-linear functional weakly separates s and t. If t is nonempty, this follows from
geometric_hahn_banach_of_nonempty_interior; if t = ∅, the zero functional works.
If A is convex with nonempty interior and x ∉ interior A, then there is a nonzero
continuous 𝕜-linear functional whose real part attains its maximum on A at x.