1 paper
Xinrui Li, Yakui Huang
Nonmonotone gradient methods generally perform better than their monotone counterparts especially on unconstrained quadratic optimization. However, the known convergence rate of th…