Hi Bili,

I've looked into this, and I think the issue is that a given oct is refined based on the numebr of particles in it, but the way that the count is done is based on the number of cells.  If you change to over_refine_factor=0 (so that the number of cells is the same as the number of octs) then you get n_p mean of about 16, which should be okay.

My check of this can be seen here, where I experimented with setting it to 2, instead of 1 (default) or 0.



On Thu, Feb 4, 2016 at 12:57 PM, Bili Dong - Gmail <qobilidop@gmail.com> wrote:
Hi all,

I'm looking at the number of particles per octree node (n_p hereafter) (a.k.a. the ('deposit', 'all_count') field) and find the results confusing. To my understanding, an octree node is refined (into 8 sub-nodes) only when n_p (of that node) > n_ref, so eventually n_p.mean (the average value of n_p) should be greater than n_ref/8. But the number I get is much smaller. When n_ref = 64, I get n_p.mean approximately 2 for all my tests.

I'm wondering if my understanding of the octree refining process if correct? And if so, how to explain the weird distribution of n_p?

Here's the link to my tests: https://nbviewer.jupyter.org/gist/qobilidop/d8f4cb85f1b465770854


yt-dev mailing list