Inspired by alternating direction method of multipliers and the idea of operator splitting, we propose a efficient algorithm for solving large-scale quadratically constrained basis pursuit. Experimental results show that the proposed algorithm can achieve 50~~100 times speedup when compared with the baseline interior point algorithm implemented in CVX.

翻译：在乘数交替方向法和操作员分裂概念的启发下,我们提出了一个高效的算法来解决大规模四级制约基础追踪问题。 实验结果表明,与CVX实施的基线内点算法相比,拟议的算法可以实现50-100倍的加速。