"transfer matrix has 0 elements" in TBtrans

[AsymmetryChou]

Describe the issue Dear developers, Hello! I encounter a problem when run TBtrans for calculating TB model of an carbon chain with 12 atoms, where 4 for central part and 4 for left/right electrode resepctively.

• The chain along Z-direction are placed with 0.16 nm between nearest atoms.
• I use sisl to finish the structure-read and electronic structure set up. (read from VASP structure file and set onsite/hopping in sisl) As I'm considering a toy model, I take the nearest interation approximation on hopping element.
• After write nc files, I run TBtrans and have the problem:Elec: transfer matrix has 0 elements. The self-energy cannot be calculated. Please check your electrode electronic structure.
• I have checked the periodic condition for leads: In RUN.fdf, for left -A3 and for right +A3. Hamiltonian{non-zero: 12, orthogonal: True, Spin{unpolarized, kind=f}, Geometry{na: 4, no: 4, Atoms{species: 1, Atom{C, Z: 6, mass(au): 12.01070, maxR: -1.00000, Orbital{R: -1.00000, q0: 0.0} }: 4, }, maxR: -1.00000, Lattice{nsc: [1 1 3], origin=0.0000, 0.0000, 0.0000, A=[10.0000, 0.0000, 0.0000], B=[0.0000, 10.0000, 0.0000], C=[0.0000, 0.0000, 6.4000], bc=[Periodic, Periodic, Periodic] } } }

I have no idea where the problem is. The lead( with 4 atoms) Hamiltonian elements are: [[-0.50120997 0.02240228 0. 0.02240228] [ 0.02240228 -0.50120997 0.02240228 0. ] [ 0. 0.02240228 -0.50120997 0.02240228] [ 0.02240228 0. 0.02240228 -0.50120997]]

Thank you very much!

Version details Run the below code and add to issue (if an issue is relevant for the issue): 3.9.18 (main, Sep 11 2023, 13:41:44) [GCC 11.2.0] 0.14.3.dev54+g83feb814c

[AsymmetryChou]

The RUN.fdf is :

TBT.HS chain.nc TBT.k [50 50 1] TBT.DOS.Gf true TBT.DOS.A true %block TBT.Elec.Left HS lead_L.nc semi-inf-direction -A3 electrode-position 1 %endblock TBT.Elec.Left %block TBT.Elec.Right HS lead_R.nc semi-inf-direction +A3 electrode-position end -1 %endblock TBT.Elec.Right

[AsymmetryChou]

I think I have solved this problem. The mistake above may result from ignoring hopping elements between unit cells. The interesting thing now is that: the Lead Hamiltonina is ( read by sisl.Hamiltonian.H[i,j] ,i/j range from 0 to 3) [[-0.50120997 0.02240228 0. 0. ] [ 0.02240228 -0.50120997 0.02240228 0. ] [ 0. 0.02240228 -0.50120997 0.02240228] [ 0. 0. 0.02240228 -0.50120997]] which contains 10 non-zero elements. However, if I print out the Lead Hamiltonian.nnz, it shows 12 non-zero element. So it hides the 2 elements corresponding connection between nearest unit cells. That's the reason I encountered the above mistake.

[zerothi]

Seems like you solved the problem your self?

If anything, it seems you are building the structure erroneously, something about the couplings and nsc. If you want help, please post the code that creates the Hamiltonian. Also, please put your code (and fdf-inputs) in triple back ticks to get code blocks: ```