Failure to run Two-Site Model with fully diagonal block structure

[Wentzell]

The following test for a two-site model with density-density interactions produces results different from the case where the full block-structure is not assumed (i.e. only spin block structure). In fact, the average perturbation order is returned as 0 for this test...

from pytriqs.gf import *
from pytriqs.operators import *
from triqs_ctint import Solver

######## physical parameters ########
U = 1.0
V = 0.3
mu = U/4.0
beta = 10.0
eps = 0.2

# --------- set up static interactions and the block structure ---------
block_names = ['up_0','up_1','dn_0','dn_1']
gf_struct = [(bl, [0]) for bl in block_names]
h_int =  U * ( n('up_0',0)*n('dn_0',0) + n('up_1',0)*n('dn_1',0) ) \
       + V * ( n('up_0',0)*n('up_1',0) + n('dn_0',0)*n('dn_1',0) + \
               n('up_0',0)*n('dn_1',0) + n('dn_0',0)*n('up_1',0) )

# --------- Construct the ctint solver ----------
S = Solver(beta = beta,
               gf_struct = gf_struct,
               n_iw = 200,
               n_tau = 100001)

# --------- Initialize the non-interacting Green's function ----------
for bl, g_bl in S.G0_iw: g_bl << inverse(iOmega_n + mu - inverse(iOmega_n - eps));

# --------- Solve! ----------
S.solve(h_int=h_int,
        n_cycles = 1000,
        length_cycle = 50,
        n_warmup_cycles = 100)