Hi,
Can anyone please describe what the best strategy is for finding the lowest energy state for each Sz sector? For example, I am interested in a finite 2d triangular lattice with NN interaction with Heisenberg Antiferromagnetic coupling. I guess conserving the quantum number will increase the additional convergence issue during the DMRG run.
Strategy for Finding the Lowest Energy State in each Sz sectors
Re: Strategy for Finding the Lowest Energy State in each Sz sectors
To get the ground state energy within each sector, the strategy needs to be to just run DMRG for each sector separately, i.e. use charge conservation and run DMRG with an initial state in that sector.
In general, turning on charge conservation should not cause any convergence issues with DMRG and essentially be equivalent to the case without charge conservation - mathematically, you can diagonalize the Hamiltonian within each sector separately, and with charge conservation you just fix in which sector you work. The only case where this is problematic is if you have spontaneous symmetry breaking between different charge sectors, then the charge conservation can force you to write superpositions ("cat states") of different (low-entangled) states that you would naturally use otherwise.
In general, turning on charge conservation should not cause any convergence issues with DMRG and essentially be equivalent to the case without charge conservation - mathematically, you can diagonalize the Hamiltonian within each sector separately, and with charge conservation you just fix in which sector you work. The only case where this is problematic is if you have spontaneous symmetry breaking between different charge sectors, then the charge conservation can force you to write superpositions ("cat states") of different (low-entangled) states that you would naturally use otherwise.