Hi,

was this random result obtained by ALPS or by your code? Could you please send us the input files if it was with ALPS? Since the rung correlation function is actually part of the Hamiltonian I don't think that the energy is converged in your case.

Matthias

On 12 Apr 2011, at 03:40, 孙照宇 wrote:

Dear ALPS users and developers:

I want to study a spin ladder with four-site interaction. [ Phys. Rev. B 67, 100409(R) (2003), http://prb.aps.org/abstract/PRB/v67/i10/e100409 ]

As ALPS 2.0 do not support 4-site terms, I code DMRG myself. For small size system(L=2,4,6,8,10) and the four-site interaction is zero, I have checked my DMRG code by comparing the obtained ground-state energy with that obtained by ED in ALPS. The two are fairly consistent.

Then I move on to larger systems and four-body term are included. However, I am confused by the DMRG results: in the staggered dimer state of the system(see Phys. Rev. B 67, 100409(R) (2003), Fig. 2), when the ground-state of the ladder has converged, I check the two-site correlations, and find that the correlation function [ G_leg ] between nearest-neighboring two sites on the leg converges very well, meanwhile the correlation function [ G_rung ] between the two sites on the same rung does not converge at all ( G_rung seems to be random with the increase of size L, L = 100 ).

I wonder is this a reasonable result ( is it possible that G_rung becomes undefined or because of some other physical reason, leading to a random average value for G_rung ) ? Or all the pairwise correlations in the system should always be converged after the gound-state energy converges ? Any suggestion would be appreciated.

sunzhaoyu2000@gmail.com