A high performance, open source, parallel MMF library written in C++.
Authors: Risi Kondor, Nedelina Teneva and Pramod K. Mudrakarta.
Documentation: [soon]
Gallery: [gallery]
BibTeX entry: [bib]
Help/support email: pMMF.library at gmail.com
Compiling pMMF requires a C++11 compatible compiler such as clang. GCC compatibility is not guaranteed, since GCC's C++11 support is still experimental. pMMF makes extensive use of the C++ Standard Template Library, which is bundled with most C++ compilers.
Certain features of pMMF also require the following:
Eigen (for matrix inversion and certain tests)
MatIO (for MATLAB format input/output)
- What is MMF used for?
- Matrix compression/sketching by providing a good quality low dimensional approximation to huge matrices, especially those arising in machine learning problems.
- Preconditioning to accelerate the solution of large linear systems.
- Sparse approximation by providing a multiresolution (wavelet) basis for A.
- Data analysis by helping to uncover the structure of A itself, for example, when A is the adjacency matrix of a complex network.
- What is Parallel MMF (pMMF)?
pMMF is a fast concurrent algorithm for finding MMF factorizations. Empirially, on many classes of naturally occurring sparse matrices, pMMF runs in time close to linear in n.
- Where can I read more about it?
- R. Kondor, N. Teneva and V. Garg: Multiresolution Matrix Factorization (ICML 2014) [supplement] [video]
- R. Kondor, N. Teneva and P. K. Mudrakarta: Parallel MMF: a Multiresolution Approach to Matrix Computation (preprint, July 2015)
pMMF is shared under the terms of the GNU Public License V.3.0. The copyright to pMMF is retained by the authors, Risi Kondor, Nedelina Teneva and Pramod K. Mudrakarta, and the authors reserve the right to separately license the code, in part or in whole, to commercial entities.