You can also store vectors or matrices in a split-up fashion as separate rows in a table, which is particularly useful if they're sparse. I've handled huge sparse matrix expressions (add, subtract, multiply, transpose) that way, cause numpy couldn't deal with them.