fedoo.problem.NonLinear.to_start

NonLinear.to_start()

Restore the state of the last converged increment.

Called when the Newton-Raphson algorithm fails to converge: the displacement increment, the minimal number of sub-iterations, the load-factor increment and the initial error are reset, then the assembly and the constraints are brought back to their committed state. The tangent matrix is re-assembled at the restored state, so that the next attempt (with a smaller time increment, see nlsolve()) starts its elastic prediction from the same matrix as the increment that has just been cancelled.