Given a convex function , its convex conjugate gives the (negative) intercept of the tangent to of slope :
#figure
This gives an alternate way to describe , not as a series of points drawing its curve, but as series of lines hugging its curve.
Duality
Put another way, is minimized under the constraint that
which shows by symmetry that is also ’s convex conjugate: indeed, there is also no way to make smaller without lowering one of its tangents, which would increase , so is also minimized under constraint .
Derivatives
For a moment, suppose that is derivable and strictly convex, which means exists and is smoothly increasing.
Consider the tangent of slope , and suppose it touches at (which means ). Then intuitively, if we slightly increase the slope , gliding the tangent further along , the intercept will decrease at a rate proportional to since it is a distance away, meaning that . That is, the derivatives of and are inverses of each other, and in particular this shows that is also convex.
#figure
Slightly more formally, since is smooth, we have for close to , so for a small slope increase , if we take the point such that , we have
which shows that as desired.