Variation of Initial shear tractions since simulating spontaneous earthquakes

I’m trying to simulating spontaneous rupture by RateStateAgeing, using my own 3D geometry model, getting the initial tractions from my coseismic solution according to 2D subduction example, and then running. Exactly, I have my simulating solution, but the initial shear tractions of the first timestep in my solution is different from which I have forced in TractPerturbation in .cfg files(getting from my coseismic solution). To be specific, the all initial shear tractions I have forced are positive, but that of some vertices are negtive in the first timestep in my solution. I don’t know why it is, I think the first timestep initial tractions shound be interpolated, so it should be same as the initial tractions forced in TractPerturbation.
Thank you for your reply!

Please explain in more detail what you are doing and what you are seeing. What are your boundary conditions? Do you see any deformation at the first time step, which could be causing the tractions to be different than the initial tractions? Do the linear and nonlinear solvers converge? I recommend supplying a diagram that illustrates what you are trying to simulate along with plots of the initial traction field you are imposing and what you are getting in the output.