On the efficiency of data representation on the modeling and characterization of complex networks

► Sparse representations of networks are found to be significantly more efficient than the adjacency matrix representations for many problems if the network is not dense. ► For networks where the average degree is close to the number of nodes, sparse representations are slower. ► For some tasks, like the computation of the clustering coefficients, the sparse representation can be slower than the adjacency matrix representation even for relatively sparse networks. ► When using adjacency matrices, the use of representations for the elements of the matrix with low memory requirements can have a significant impact on performance.