Dear all,
I try to use mps_optim for the 1D Bose-Hubbard model in the presence of a trapping potential. I use the latest SVN version (from last week), the latticel model "inhomogeneous chain lattice" and a homemade model "harm boson Hubbard" with the site term: -mu#*n(i)+U#*n(i)*(n(i)-1)/2+K#*n(i)*(x-0.5*(L-1))^2 which differs simply from the usual Bose-Hubbard model by the addition of the last term K#*n(i)*(x-0.5*(L-1))^2. Such a model works with the worm QMC algorithm, but fails with mps_optim with the error message: Exception thrown: Cannot evaluate x I tried to replace "x" by "i" without success.
How should I proceed to have mps_optim work for an inhomogeneous system? More generally, I find it always **VERY** difficult to obtain the information on what should work for a specific algorithm/model in ALPS and what should not work. The only solution I found is trial and error, which is a bit frustrating.
Thanks in advance for your answer(s).
Dominique