Normalised Plummer potential gradient with and without a tapering function. In the SVSL method, the introduction of a tapering function (Eq. (29)) avoids the divergence of the Plummer potential. At the same time, the low tapering length, d_s(= 3/127 ≃ 0.024) ensures that most of the SG contribution is not lost over short distances (d_g ≤ 1.5). Under the CSL approach (red dotted), the gradient is cancelled out at the singularity, although most of the short distance contribution to the SG is lost.