Local convexification of the Lagrangian function in nonconvex optimization (Q1973484)

The authors propose an approach to locally convexify the Lagrangian function of a nonconvex optimization problem. They show that the local convexity condition in the local duality theorem can be achieved at a local optimal solution of the nonconvex primal problem after adopting some suitable convexification transformation on this problem. Linear independence of the gradients of the active constraints and the second-order sufficiency condition at the local optimal solution are assumed. Under these assumptions, the authors show that a \(p\)-power transformation yields the desired convexification of the primal problem. Two numerical examples are presented.