# Closed form for Lagrange dual

Can Lagrange dual always be computed in closed form? Can you give me a simple example where the dual is not analytically computable?

-