GPS资料同化中一种快速优化算法

GPS资料同化最终归结为一个大规模的无约束优化问题。由于问题的维数很高，使得寻找到一种快速且节约存储的优化算法成为GPS资料同化能否满足在业务数值天气预报上的要求的关键。提出了一种新的优化算法。该方法把L-BFGS和HFN两种方法动态的结合起来，同时利用有限差分技术和截去技术，使算法在存储和计算量适合当前设备的前提下，较大的提高了算法的收敛速度。数值试验表明，该方法的计算效率与L-BFGS方法相比有十分明显的改善。

A rapid optimization algorithm for GPS data assimilation

Global Positioning System (GPS) meteorology data variational assimilation can be reduced to theproblem of a large-scale unconstrained optimization. Because the dimension of this problem is too large,most optimal algorithms cannot be performed. In order to make GPS/MET data assimilation able tosatisfy the demand of numerical weather prediction, finding an algorithm with a great convergence rateof iteration will be the most important thing. A new method is presented that dynamically combines thelimited memory BFGS (L-BFGS) method with the Hessian-free Newton(HFN) method, and it has a goodrate of convergence in iteration. The numerical tests indicate that the computational efficiency of themethod is better than the L-BFGS and HFN methods.