The jeans test currently highlights the inaccuracy of tyradiational tree based methods when the gravitational forces nearly cancel. We need to implement a relative opening criterion that will work robustly for problems of this nature, a la GADGET2.
At the moment there seems to be a serious problem with periodic gravity in the master branch. Even when all of the nodes are opened the result seems to be junk...
The jeans test currently highlights the inaccuracy of tyradiational tree based methods when the gravitational forces nearly cancel. We need to implement a relative opening criterion that will work robustly for problems of this nature, a la GADGET2.