A gallery of large graphs

Graph visualization is a way to discover and visualize structures in complex relations. What sort of structures are people who do large scale computation studying? We can get a glimpse by visualizing the thousands of sparse matrices submitted to the University of Florida Sparse Matrix collection. The resulting gallery contains the drawing of graphs as represented by 2218 sparse matrices in this collection. Each of these sparse matrices (for rectangular matrix, an augmented matrix is formed first) is viewed as the adjacency matrix of an undirected graph, and is laid out by a multilevel graph drawing algorithm. If the graph is disconnected, then the largest connected component is drawn. The largest graph (Schenk@nlpkkt240) has 27,993,600 vertices and 366,327,376 edges. A simple coloring scheme is used: if the matrix has real entries, coloring is based on the entry value, otherwise it is based on the edge length.

GHS_psdef/cvxbqp1	GHS_psdef/finance256	GHS_psdef/ford1	GHS_psdef/ford2	GHS_psdef/gridgena
GHS_psdef/hood	GHS_psdef/inline_1	GHS_psdef/jnlbrng1	GHS_psdef/ldoor	GHS_psdef/minsurfo
GHS_psdef/obstclae	GHS_psdef/oilpan	GHS_psdef/opt1	GHS_psdef/pds10	GHS_psdef/pwt
GHS_psdef/ramage02	GHS_psdef/s3dkq4m2	GHS_psdef/s3dkt3m2	GHS_psdef/srb1	GHS_psdef/torsion1