Consider the optimization problem P: minx∈Rdf(x) where ∀i∈I,ci(x)≥0∧∀j∈J,hj(x)=0 The corresponding Lagrangian is L(x,λ,μ)=f(x)−λTc(x)−μTh(x) Define g as g(λ,μ)=minx∈RdL(x,λ,μ) Let D be this optimization problem: maxλ,μg(λ,μ) where g(λ,μ)≠−∞∧λ≥0 Then D is said to be the dual of P.