On 18 Nov 2013, at 21:46, Rafał <arymanus@gmail.com> wrote:

Dear Matthias,

Thank you for your reply. I have one more question about DMRG.

I try to calculate a gap of xxz spin model (Jz != Jxy) for different values of Jz and Jxy. I think that it will be more safe to assume that I do not know in what Sz sector search for a ground and excited states. 

Is it in this case better option to run simulation over all possible Sz by hand or to comment a line 
#'Sz_total' : Sz.
which should force alps to  work in the grand canonical as it was told in the first tutorial on the webpage?

I tried this second approach for the ising model (Jxy = 0, Jz = 1) but what I get was just a duplicated ground state without a correct excited state.

That is because there are two ground states.

What you want to do is to instead run with Sz_total=0 and Sz_total=1 since by spin gap we usually mean the gap between the singlet and triplet sector.