1: \begin{abstract}%

2: 

3: An efficient gradient-based method to solve the volume constrained

4: topology optimization problems is presented. Each iterate of this

5: algorithm is obtained by the projection of a Barzilai-Borwein step

6: onto the feasible set consisting of box and one linear constraints

7: (volume constraint). To ensure the global convergence, an adaptive

8: nonmonotone line search is performed along the direction that is

9: given by the current and projection point. The adaptive cyclic

10: reuse of the Barzilai-Borwein step is applied as the initial

11: stepsize. The minimum memory requirement, the guaranteed

12: convergence property, and almost only one function and gradient

13: evaluations per iteration make this new method very attractive

14: within common alternative methods to solve large-scale optimal

15: design problems. Efficiency and feasibility of the presented

16: method are supported by numerical experiments. \\

17: 

18: %===============================================================

19: 

20: \noindent{\bf Keywords and phrases. }%

21: Barzilai-Borwein step-size, distributed parameter identification,

22: large-scale topology optimization, method of moving asymptotic

23: (MMA), nonmonotone line search, volume constraint.

24: 

25: %===============================================================

26: 

27: \end{abstract}%