GCMMA-MMA-code

Description of the algorithms used in the implementations of MMA and GCMMA in Matlab. Files for both MMA and GCMMA: ============================ subsolv.m kktcheck.m toy1.m toy2.m subsolv.m ========= The function subsolv.m makes an attempt to solve either the MMA subproblem generated by mmasub.m, or the GCMMA subproblem generated by gcmmasub.m. A straight-forward primal-dual interior-point method is used. kktcheck.m ========== The function kktcheck.m calculates the left hand sides of the KKT conditions for the optimization problem defined by the user. toy1.m and toy2.m ================= Define, together with mmatoyinit.m or gctoyinit.m, the users problem. ---------------------------------------------------------------- Files for MMA only: ================== mmasub.m mmatoymain.m mmatoyinit.m mmatoyresults mmasub.m ======== The function mmasub.m generates the MMA subproblem for the current iteration, and calls the function subsolv.m mmatoymain.m ============ The file mmatoymain.m makes an attempt to solve the users optimization problem which is defined by the files mmatoyinit.m and toy2.m. In each MMA iteration, mmatoymain.m calls toy2.m and mmasub.m. ---------------------------------------------------------------- Files for GCMMA only: ==================== gcmmasub.m asymp.m concheck.m raaupdate.m gctoymain.m gctoyinit.m gctoyresults gcmmasub.m ========== The function gcmmasub.m generates the GCMMA subproblem for the current outer/inner iteration, and calls the function subsolv.m. asymp.m =========== The function asymp.m calculates values on the parameters raa0, raa, low and upp in the beginning of each outer GCMMA iteration. concheck.m =========== The function concheck.m checks if the current GCMMA approximations are sufficiently conservative. raaupdate.m =========== The function raaupdate.m updates the parameters raa0 and raa in each inner iteration. gctoymain.m =========== The file gctoymain.m makes an attempt to solve the users optimization pr