一类新的自适应非单调谱投影梯度法

In this paper, we give a new class of adaptive nonmonotone spectral projected gradient technique for convex constrained optimization. The main idea of our method is that we introduce a self-adaptive weight parameter which makes the algorithm can adjust the nonmonotone strategy automatically. We obtain the global convergence of the proposed algorithm under certain conditions and the numerical tests show the algorithm can reduce the dependence on the nonmonotone parameter to some extend.