Skip to main content
. 2018 Jul 13;34(24):4266–4273. doi: 10.1093/bioinformatics/bty514

Fig. 2.

Fig. 2.

Illustration of the computation of an optimal scaling parameter s^i for Laplace noise. (A) Objective function J for fixed θ and different values of σi, showing that the kinks, i.e. the points of non-differentiability, indicated by the dashed lines are independent of σi. (B) Derivative of the objective function with respect to the scaling parameter which is not defined at the kinks. The light red and dark red lines indicate the computed scaling parameter and the true optimal scaling parameter, respectively