Bulk vs boundary twist

[ALF-Import-Bot]

In GitLab by @joaogci on Jul 28, 2025, 13:10 UTC:

Dear all,

I have encountered an issue when introducing a twist in the bulk (Bulk = .T.) or in the boundary (Bulk = .F.). I found this while debugging an Hamiltonian in which the limit of an interaction gives a twist in the boundary conditions. I use the honeycomb lattice and just consider a system of nearest-neighbour hopping fermions. Due to gauge invariance, both methods of introducing the twist should result in the same energy, for instance. Furthermore it the twist should be periodic in Phi_x.

Attached you can find the file where I include the ham_int and sampling functions in my Hamiltonian. The hoping vertices are defined in op_V as I use a type 3 field. In the case where I ran I manually set the field to zero at all operators and time slices. I also attach the parameter file.

The results for the Kinetic energy as follows:

• Bulk = .F.

• Phix = 0 -1.56527

• Phix = 0.5 -1.57430

• Phix = 1 -1.56527

• Bulk = .T.

• Phix = 0 -1.56527

• Phix = 0.5 -1.52613

• Phix = 1 -1.35992

undefined
undefined
undefined

Migrated from GitLab: https://git.physik.uni-wuerzburg.de/ALF/ALF/-/issues/318

[ALF-Import-Bot]

In GitLab by @fassaad on Oct 28, 2025, 09:14 UTC:

@joaogci
For the square lattice Phi_x = 0.5 I get, for the Kinetic energy:

2 (-0.58864643576962365E+002,-0.28257607152726080E-013) 0.10000000000000000E+001 Bulk = T
2 (-0.58864643576962166E+002, 0.18032621711173783E-031) 0.10000000000000000E+001 Bulk = F

This is the result for the Honeycomb lattice:

2 (-0.11256810434638712E+003, 0.23052535340023132E-028) 0.10000000000000000E+001 Bulk = F
2 (-0.11256810434638723E+003,-0.80839878654770590E-013) 0.10000000000000000E+001 Bulk = T

I use the this Hamiltonian https://git.physik.uni-wuerzburg.de/ALF/ALF/-/blob/master/Prog/Hamiltonians/Hamiltonian_Hubbard_smod.F90?ref_type=heads

Here the kinetic energy is computed with Call Predefined_Hoppings_Compute_Kin(Hopping_Matrix,List,Invlist, Latt, Latt_unit, GRC, ZKin) Of course the flux is included in the computation of the Hamiltonian.

Hence it seems that I cannot reproduce the bug.

[ALF-Import-Bot]

In GitLab by @joaogci on Oct 31, 2025, 06:55 UTC:

@fassaad I get the same values as you with the default Hamiltonian, so I cannot reproduce the error anymore. When I had this issue I was also computing the kinetic energy with the mentioned function, but in my private branch with the matrix fields (with the fields set to zero, essentially just hopping). So maybe it is a problem from there.