Hi, recently, I tried to do my research on Ubuntu with oofem, and it is convinient to run oofem on Ubuntu. However, I found that it is easier to diverge on Ubuntu with PETSc than it on Windows. I have searched some information about it in the forum.

solution starts to diverge. This is caused by several things: 1) your mesh is really coarse and steps are too big. Also the nonlocal model requires at least several neighboring integration points to be within the the nonlocal averaging zone and from this point of view the grid is again coarse (and may be that the value you are using is not physically based, since the averaging radius is closely related to the width of fracture process zone). The fact that you have obtained convergence for larger steps was rather coincidence.
So my suggestion is to use finer grid with smaller steps.

The number of meshes is 1000,000, so I think it is refine enough. I still don't understand how to apply the static forces in several steps. The number of nodal force is big, several hundred. But isn't it applied at same time?