Let . Maximizing a linear functional over this line segment occurs at an endpoint, so
This identifies as a support function. Its convex conjugate is the indicator functional : if , then for every , with equality at zero; if , strict separation and positive scaling of the separating vector make the supremum infinite.
For a convex function, equality in the Fenchel–Young inequality characterizes its subdifferential. Hence
This is support-function subgradients as exposed faces. In particular a maximizing really is a subgradient, since for every .
Writing , with , now gives
If , the last line is the singleton , so the formula includes that case.
Support function Created 2026-09-28 Updated 2026-11-05
The support function of a nonempty subset of a finite-dimensional real inner-product space is
It gives the largest value of a linear objective in direction , allowing when the set is unbounded in that direction. As a supremum of linear functions, it is convex and lower semicontinuous; it satisfies for and .
For the Euclidean unit ball, the Cauchy-Schwarz inequality gives , attained by when . For the probability simplex, the weighted-average bound gives , attained at a maximizing coordinate vector. Thus many familiar convex objectives are support functions.
Let be the indicator functional of a constraint set, equal to zero on and outside it. The convex conjugate definition gives . If is closed and convex, the Fenchel-Moreau theorem then gives . This uses the extended-real constraint indicator, rather than the zero-one indicator function used in measure theory.[ref]
The support-function subgradients as exposed faces identifies the maximizing vectors, linking sensitivity of the support function to the geometry of .