The aim of this paper is to present an improvement of one well-known method for d = 3. In the original paper two algorithms were presented, one for d = 3 and another (Chinese remainder sieve method) that was modifiable for arbitrary d.
In its basic form, the Chinese remainder sieve method was always better than the explicit algorithm for d = 3. In our proposed form, the modified algorithm for d = 3 is more efficient for some small n, and it also pushes the lower bound from which an efficient algorithm exists