reduce_montgomery
out->coeff[i] = reduce_montgomery((uint64_t)lhs->coeff[i] * (uint64_t)rhs->coeff[i]);
uint32_t t_odd = reduce_montgomery((uint64_t)z_step_root
p->coeff[j + offset] = reduce_montgomery((uint64_t)step_root
p->coeff[i] = reduce_montgomery((uint64_t)p->coeff[i] * (uint64_t)inverse_degree_montgomery);