We introduce a new class of augmented Lagrangian function, which includes the well-known essential quadratic augmented Lagrangian as special cases. Based on this new function, we propose a multiplier algorithm, whose main feature is that the multiplier sequence does not require to be bounded. Global convergence to optimal solutions and KKT points are established, respectively.
Zhou, Jinchuan; Xu, Xiuhua; and Tang, Jingyong
"New Multiplier Algorithm for Nonlinear Programming with Inequality Constraints,"
Applied Mathematics & Information Sciences: Vol. 06
, Article 22.
Available at: https://dc.naturalspublishing.com/amis/vol06/iss2/22