I'm aware that typical techniques to store matrices in sparse form are compressed formats or maps where the key is the pair of indices and value the value of the entry in a matrix.
I was wondering if vEB trees could be used to such purpose as well. At the end of the day the dynamic set to be stored would be the pairs $(i,j)$ in lexicographic order. If $MN$ is the size of the matrix a vEB tree would allow the access to a specific entry in time $O(log(log(MN))$ which isn't bad, though the space requirement doesn't change, but maybe with some improvements something interesting could come out.
Is there some research in this direction? I quickly looked up on google doesn't seem anything specific comes out.