Which scenario is least suitable for sparse matrix representation in computer memory?

1. AAn adjacency matrix for a large network
2. BA large matrix with many zeros
3. CA diagonal matrix
4. DA matrix where most elements are nonzero