顶！求大佬帮助！

查找Fitting fully observed recursive mixed-process models with cmp   这篇文章，似乎也未给出具体的解释，请大佬指导

文中说到：  To make results easier to interpret, cmp representsΣ in“sigma-rho” form, that is, with a standard deviation (σ) parameter for each error and a correlation coefficient (ρ) for each pair. Because these parameters are bounded, they are transformed onto an unbounded scale by using the logarithm of theσ’s and the arc-hyperbolic tangents (inverse S-curve transforms) of theρ’s. Fitting with lnsig and atanhrho parameters eliminates the possibility that in the course of its search, ml will submit impossible trial values for the parameters, such as a negative value for a σ.

机翻：为了使结果更易于解释，cmp 以“sigma-rho”形式表示 Σ，即每个误差都有一个标准偏差 (σ) 参数和每一对的相关系数 (ρ)。因为这些参数是有界的，所以通过使用 σ 的对数和 ρ 的反双曲正切（反 S 曲线变换）将它们转换为无界形式。拟合 lnsig 和 atanhrho 参数消除了在搜索过程中 ml 将提交不可能的参数试验值的可能性，例如一个负值 σ 。

dddddddd

每日一逛