These matrices cover a wide spectrum of domains, include those arising from problems with underlying 2D or 3D geometry (such as structural engineering, computational fluid dynamics, model reduction, electromagnetics, semiconductor devices, thermodynamics, materials, acoustics, computer graphics/vision, robotics/kinematics, and other discretizations) and those that typically do not have such geometry (such as optimization, circuit simulation, economic and financial modeling, theoretical and quantum chemistry, chemical process simulation, mathematics and statistics, power networks, and other networks and graphs). The collection is widely used for the development and testing of graph-theoretic algorithms, graph-visualization methods, and sparse matrix algorithms, among other applications. The collection includes software for accessing and managing the collection, from MATLAB, Fortran, and C.
As of April 2009, it contains 2255 problems (some of which are sequences of dozens of matrices). The collection is an extension of the Harwell-Boeing Collection, which comprises the first 292 matrices in the UF Sparse Matrix Collection. The largest matrix in the UF collection has a dimension of almost 29 million, with 760 million nonzero entries. The matrices are available in three formats: MATLAB mat-file, Rutherford-Boeing, and Matrix Market (http://math.nist.gov/MatrixMarket). Note that the MATLAB mat-files can only be read by MATLAB 7.0 or later. The source of this collection is http://www.cise.ufl.edu/research/sparse/matrices.
References: When presenting results on these matrices, please cite the following papers:
T. A. Davis, The University of Florida Sparse Matrix Collection, ACM Transactions on Mathematical Software (submitted, 2009), also available as a tech report at http://www.cise.ufl.edu/~davis/techreports/matrices.pdf
When using matrices in the HB/ group in the collection, or for the Rutherford/Boeing format, please cite this paper:
I. S. Duff, R. G. Grimes, and J. G. Lewis, Sparse Matrix Test Problems, ACM Transactions on Mathematical Software, vol. 15, no. 1. pp. 1-14, 1989, http://doi.acm.org/10.1145/62038.62043
Data aggregated, processed and curated by Tim Davis using matrices contributed by the world academic community.
Snapshots prepared by the infochimps.org team using community curated metadata.
The images above were created by Yifan Hu, AT&T Labs.