[Diger]

Changing the underrelaxation factor for example for the energy from 1 to 0.5 does it actually change the solution how it converges and once it's converged, so after a decent amount of iterations or is it necessary to switch off the "higher order term relaxation" and (the underrelaxation factors only refer to this higher order term relaxation, right?) then iterate to convergence again.
Does this feature (higher order term relaxation) only intervene in case of divergence detection or is the new variable always calculated by phi = phi + alpha*dphi with the under relaxation factor alpha. I'm not quite sure when the variable is updated this way and when it's not. Atm I understood it such that the update is always phi=phi + dphi but in case of divergence it tries again via phi = phi + alpha*dphi. Am I right or wrong?