Correction to the computation of the norm of the damage correction in the phase-field demo
The magnitude of the damage correction between two iterations is computed error as follows:
# Solve damage d-problem
dval_old = d.vector().get_local()
problem_d.solve(d.vector())
# Residual on damage increment
dval = d.vector().get_local()
res = np.max(dval - dval_old)
In my understanding, it should be:
res = np.max(np.abs(dval - dval_old))