TY - JOUR
T1 - An augmented Lagrangian approach to non-convex SAO using diagonal quadratic approximations
AU - Groenwold, A.A.
AU - Etman, L.F.P.
AU - Kok, S.
AU - Wood, D.W.
AU - Tosserams, S.
PY - 2009
Y1 - 2009
N2 - Successful gradient-based sequential approximate optimization (SAO) algorithms in simulation-based optimization typically use convex separable approximations. Convex approximations may however not be very efficient if the true objective function and/or the constraints are concave. Using diagonal quadratic approximations, we show that non-convex approximations may indeed require significantly fewer iterations than their convex counterparts. The nonconvex subproblems are solved using an augmented Lagragian (AL) strategy, rather than the Falk-dual, which is the norm in SAO based on convex subproblems.
AB - Successful gradient-based sequential approximate optimization (SAO) algorithms in simulation-based optimization typically use convex separable approximations. Convex approximations may however not be very efficient if the true objective function and/or the constraints are concave. Using diagonal quadratic approximations, we show that non-convex approximations may indeed require significantly fewer iterations than their convex counterparts. The nonconvex subproblems are solved using an augmented Lagragian (AL) strategy, rather than the Falk-dual, which is the norm in SAO based on convex subproblems.
U2 - 10.1007/s00158-008-0304-x
DO - 10.1007/s00158-008-0304-x
M3 - Article
SN - 1615-147X
VL - 38
SP - 415
EP - 421
JO - Structural and Multidisciplinary Optimization
JF - Structural and Multidisciplinary Optimization
IS - 4
ER -