Bug Report about BoundCorr in InOutExtrapolator
I noticed that first-order BoundCorr was never used in the Three-dimensional case in the official example. And I found some bugs in this kind of case.
I found that in Function JSphCpu::InteractionInOutExtrap_Single. Line 395 in the figure, I believe a11, a12, a13 in gradvelp1 should be a21, a22, a23.
This bug can be found in _Double, and GPU's code.