Really fascinating work! In f(r) = pi(r) p(r), p(r) seems similar to Σ_{i,j} m_i m_j in classical gravity. Your formulation gives pi(r) = Σ_{i,j} f_{ij} / Σ_{i,j} m_i m_j, whereas classical gravity uses pi(r) ≈ (1/N) Σ_{i,j} (f_{ij}/m_i m_j). Is this correct?