Skip to main content
. 2017 Sep 22;7:12140. doi: 10.1038/s41598-017-11873-y

Figure 2.

Figure 2

Scaling of the estimated ID with respect to the number of points; for each distribution and for a number of points going from 20 to 25000 we harvest 200 instances of the dataset and average the resulting estimates for the ID. The test is carried out in dimension 2, 5 and 10. Panel A: Hypercube with pbc. Panel B: gaussian distribution. Panel C: Cauchy dataset. Panel D: uniform distribution on a hypersphere.