Pioneered by the Bronstein duo, GMDS is described in their many recent papers, e.g. [9,5,8,6]. As explained in prior parts, GMDS can be used for a lot more than face recognition as its application can be further generalised (e.g. to analysis of non-rigid shapes). GMDS deals with isometric embedding, where the intrinsic metric structure of some given surface can be represented using another surface, which in turns yields some correlation between these two. The numerical framework proposed enables the finding of correspondences. Measures of similarity can also be derived from the average metric distortion and the Gromov-Hausdorff distance advances matching of surfaces, using this exact same framework. In the context of our work, GMDS should be used for (dis)similarity in an objective function we ought to define.