Dear Prof. Troyer, in a former post I described some problems that I have found reproducing some results by Ogata et al. on the t-J model; they are still unsolved and I would really need to use the library for a more complicated problem. Briefly, I focus on one single case: N = 4, L = 16. (For N = 2, 6, 10.. things seem to work with periodic bc).
The reference suggests to use antiperiodic bc to have the ground state in the zero momentum sector. This should be equivalent (through a canonical transformation) to a system with complex hopping t -> t exp(i \pi / L) and periodic boundary conditions so that I can use the library in the standard way. In the python script I define two hoppings (left and right)
phi = Pi/L
tr = 'exp(' + str(phi)+'*I)' tl = 'exp(-' + str(phi)+'*I)'
I checked for spinless non-interacting fermions with complex hopping that I get the ground state energy correctly both in the case with zero flux and in the case with \pi flux. This can be done analitically and, if I consider an odd number of fermions, I found the correct result in the zero momentum sector for 3 particles in 8 sites and 5 particles in 16 sites. For the t-J with N = 4, L = 16 the ground state that I find is not in the zero momentum sector (E=-7.5655955) but it is in another sector (k = 5.4977871437) and the energy is (E=-7.84743613). Any suggestion?
Best,
--- Marco Di Liberto PhD Student
ITP (Institute for Theoretical Physics) Utrecht University