求解非凸函数极小的异步并行BFGS算法

Asynchronous Parallel BFGS Method for Solving Nonconvex Minimization

若假设可供使用的处理机具有p+q台,将其分成两组,两组处理机之间进行异步并行计算,该文提出了一种求解非凸函数极小的异步并行BFGS算法,若目标函数连续可微,且它的一阶导数是Lipschitz连续的,证明了并行拟牛顿算法是全局收敛的.

We assume that we have p ＋ q processors, which are diveded into two groups, the two groups execute in an asynchronous parallel fashion. In this paper, we present an asynchronous parallel BFGS method for solving nonconvex minimization if we assume the objective function is continuously differentiable and has Lipschitz continuous gradients, we establish global convergence of the parallel method.