Sumk chemical potential off by a factor of 2

[jkarp314]

I used Sumk for a simple tight binding model on a square in 2d with one orbital, only nearest neighbor hopping, and hopping parameter t=1. The half-bandwidth should be 4t=4. I density as a function of chemical potential, which should be 0 at -4 and 2 at 4. However, I am finding that the density is instead 0 at -8 and 2 and 8, and everything is off by a factor of 2.

The issue seems to be that I am using 2 spins, so the Green's function has 2 blocks. The code multiplies the chemical potential by the number of blocks, which it should not do.

import numpy as np
import matplotlib.pyplot as plt
from pytriqs.lattice.tight_binding import *
from pytriqs.gf import *
from pytriqs.sumk import SumkDiscreteFromLattice

BL = BravaisLattice(units = [(1,0,0) , (0,1,0) ])
t = -1.
hop= {  (1,0)  :  [[ t]],       
        (-1,0) :  [[ t]],     
        (0,1)  :  [[ t]],
        (0,-1) :  [[ t]]}
TB = TBLattice(units = [(1,0,0) , (0,1,0) ], hopping = hop)
SK = SumkDiscreteFromLattice(TB,n_points=8)
Sigma0 = GfImFreq(indices = ['up','down'], beta = 400)
SigmaB = BlockGf(name_list = [''], block_list = [Sigma0])

def Dens(mu):
    Glat = SK(SigmaB, mu = mu)
    dens = Glat.total_density()
    return dens 
	

mus = np.linspace(-10,10,51)
den = np.empty(51)
for i in range(51):
    den[i] = Dens(mus[i]).real

	
plt.plot(mus,den, '-o')
plt.grid()
plt.show()

[HugoStrand]

I think this is a bug. The hopping matrices are 1x1 but the target space of the self-energy is 2x2. The call to sumk should fail with an error here. Maybe related to prq #550?

Changing the self-energy target space gives the expected bandwidth.