Dear ALPS users, I want to run some DMRG simulations on 1D Fermi Hubbard models. I tried to work in the gran canonical ensemble, giving the chemical potential mu instead of the density and without specifying that Nup and Ndown are conserved quantities. The code is running but it produces strange results, in particular the calculated energy is varying a lot within the same sweep and the error oscillates from 10-15 to 10-2. Any ideas why ? Puzzled by that, I tried to do the same for the Bose Hubbard model, but there everything works perfectly. Can you help me on that ? My best regards, Giuliano