a note about density thresholds

Post date: Oct 09, 2013 3:2:10 PM

We would like to thank Mark Neyrinck, Seshadri Nadathur, and Shaun Hotchkiss for pointing out an inconsistency regarding density thresholds between the text of our Sutter et al. (2012) paper and the actual implementation. In the text, we state that we impose an overall density threshold to limit voids to less than 0.2 the mean density (rho_mean).

What is actually implemented is a 0.2*rho_mean criterion for merging zones (here, a "zone" is a group of Voronoi cells sharing a common basin) together in the process of creating larger parent voids, as described in Neryrinck (2008). Zones are only added to a void if the minimum density in that zone is < 0.2*rho_mean. This corresponds to a value of 0.2 (suitably adjusted to account for the presence of boundary particles) as the last argument to the jovoz code. Please see the ZOBOV manual page for more details.

Thus, voids presented in our catalog will be overdense on average, since the watershed algorithm naturally includes high-density edge galaxies in the void definition. Also, even the minimum density of a void can be greater than 0.2*rho_mean, since voids can (and do) consist of only a single zone, and this criterion is not applied until the zone-merging step. This, again, is discussed in Neyrinck (2008). However, as we discuss in Sutter et al. (2012), the centers of our identified voids are still underdense.

This becomes much more severe as the void size approaches the mean particle separation, since small voids tend to be surrounded by very large overcompensated regions. The philosophy of this catalog is to produce as many voids as possible and allow users to make their own cuts as they see fit. As we have before, we strongly urge caution in the use of the smallest reported voids!

We apologize for this misunderstanding, and look forward to further improvements of our catalog.