Internal vs Cartesian coordinates

As we commented, the usage of internal coordinates implies some matrix transformations between internal and Cartesian coordinates that become a computational bottleneck in big systems. Some work has been devoted to overcome this computational drawback through linear scaling techniques [163,164,161,165]. However we must take into account that even in internal coordinates, the optimization of a big system will require a large amount of steps to converge. Moreover, in flat energy surfaces such as biopolymers, the improvement on the usage of internal coordinates may be concealed under the long iterative process and the trade-off may not be  favorable.