Inertial projected gradient method for large-scale topology optimization

Abstract We present an inertial projected gradient method for solving large-scale topology optimization problems. consider the compliance minimization problem, heat conduction problem and compliant mechanism of continua. use to efficiently treat linear constraints these Also, techniques are used accelerate convergence method. adaptive step size policy further reduce computational cost. The proposed has a global property. By numerical experiments, we show that converges fast point satisfying first-order optimality condition with high accuracy compared existing methods. low cost per iteration, is thus effective in problem.