A footnote in this paper (presented at the IMF workshop last month) notes that it uses an extended version of the SOR (maybe OSR?) routine in Dynare that 'allows for constrained optimisation' and produces efficent frontiers.
I am guessing this allows a range of loss function weights to be set (to compute the full efficient frontier?). This sounded very interesting and I wondered if it was going to be added to the main Dynare package at some point.
cheers
David