L-BFGS算法比较适合在大规模的数值计算中，具备牛顿法收敛速度快的特点，但不需要牛顿法那样存储Hesse矩阵，因此节省了大量的空间以及计算资源。-Limited-memory BFGS (L-BFGS or LM-BFGS) is an optimization algorithm in the family of quasi-Newton methods that approximates the Broyden–Fletcher–Goldfarb–Shanno (BFGS) algorit

针对数据处理中，非光滑约束方程组的求解不具有全局收敛性的问题，结合光滑化技术设计了Levenberg Marquardt算法。(Aiming at the problem that the solution of nonsmooth constraint equations does not have global convergence in data processing, a Levenberg Marquardt algorithm is designed by combining sm