Interior Point Methodology
摘要
The introduction in 1984 of interior point methods (IPM) meant a sort of revolution. Today, most optimization solvers (software) include IPM, also MATLAB optimization-related functions. The chapter gradually introduces IPM, starting from some preliminaries (barrier functions, etc.) and an initial overview. Then, IPM is applied to linear programming in two ways: the primal methodology and the primal-dual methodology (in which we introduce the Mehrotra’s predictor-corrector formulation). Then, IPM is applied to quadratic programming. Next two sections use IPM for cone programming and for semidefinite programming. The end of the chapter comments some further study extensions for nonlinear programming.