I try to calculate the green’s function with pylith v2.2.2. I revised the example configuration file in
3d/subduction/step07a. I export the traction change on fault surface instead of surface deformation. The fault surface in my model contains 9844 nodes, but some of them belong to fault edge, so the final size of the traction change matrix is 3×9844×9583. However, I found that the traction change matrix is not in the order of the vertices matrix. Could you please tell me how to know what order this matrix is in?
Another question is about the amplitude of traction change. I used the slip impulse of 1 m, and the largest value in the traction change could reach to 100 MPa. Is it reasonable for such a large value of traction change?
The order of the points in the fault “data” matches the fault “mesh” (geometry/vertices and topology/cells).
The traction changes adjacent to a slip impulse can be large; the exact value depends on the geometry of the fault surface and size of the cells.
Thanks for answering. The size of the traction chage matrix is 3×9844×9583, the 9844 part is in the order of the
geometry/vertices, but the 9583 part is not the same size as the
geometry/vertices and it seems to be randomly ordered.
9583 is the total number of slip impulses. They are not randomly ordered, but ordered based on the processor and degree of freedom number. If the mesh is reordered at the beginning of the simulation (default), then the vertices are numbered to be close in space.
The fault output file for a Green’s function calculation includes the amplitude at each point on the fault for each impulse. This allows you to associate the location of the impulse with each vertex on the fault. In PyLith v2.2, see
examples/2d/greensfns/invert_slip.py for Python code on how we associate slip impulses with their response. In PyLith v3.0, the corresponding file is
Thanks a lot for helping.