The project was a result of my Dual Degree Project at IIT Madras with Prof. S. Sundar. This library takes a sparse matrix as input and constructs a Hierarchical Matrix representation for the matrix. The library uses graph clustering algorithms to find sub-matrices inside the matrix which can be represented in compressed form using reduced-SVD. The library is based on algebraic construction process and uses no geometric information, whatsoever. This process has been described in this paper. Although the library gives out a correct H-matrix, it is slightly slow compared to other highly-optimized libaries avaliable on internet such as HLibPro.
Note: the library uses a supporting library called Eigen, to handle all manipulations related to linear algebra. The documentation for the library is inside docs/index.html.
For more info look here.