Sparse Vector / Matrix Product

Requirements Acceptance criteria: Use less space than a dense representation: storage must be O(nnz), rather than $$O(R \times C)$$. dot must run in O(nnz1 + nnz2). Alternatively, if the larger operand is backed by a hashmap, iterate the smaller one for O(min(nnz1, nnz2)) work. matmul must have complexity O(nnzA × avg_nnz_per_col_B), not $$O(R \times K \times C)$$. Notes The stated requirement is that the approach must not be $$O(M \times N)$$; a dense multiplication…