This version has been tested on Matlab, Julia and C++. (Its python interface has not been tested.)
The major update is about the container classes.

*) Riemannian stochastic gradient descent (RSGD), Riemannian ADAM (RADAM), and Riemannian AMSGRAD (RAMSGRAD) types of of algorithms are added. These algorithms are implemented by Yihui Huang (Xiamen University).

*) Riemannian stochastic variance reduction gradient (RSVRG), stochastic variance reduction LRBFGS (SVRLRBFGS), and stochastic variance reduction LRBroyden family (SVRLRBroydenFamily) algorithms are added. These algorithms are implemented by Shuguang Zhang (Florida State University).

*) A Riemannian proximal gradient and its accelerated version are added (IRPG and IARPG). 

*) Three problems including PCA on Grassmannian (GrassPCA), Matrix completion on Grassmannian (GrassMatCompletion), and Poincare ball embedding (PoincareEmbeddings) are added. The two problems of soft ICA on the Stiefel manifold and Karcher mean on SPD manifold are modified. These four problems can be solved by Riemannian stochastic types of algorithms.

*) The current version is tested on Mac, Windows, and Ubuntu with C++ environment and Matlab.