Hi developers!
I’am confused by the statement in Manual Siesta4.1.5 about the Kpoint convergency. It tells me that
Once TranSIESTA is completed one should also converge the number of k points for TBtrans.
Note that k point sampling in TBtrans should generally be much denser but always
fulfill $N_k^{TS} \geq N_k^{tbt}$.
My questions are:
-
If TranSIESTA’s kpoints have resulted in convergent output, it means that the output Hamiltonian and eletron Density Matrix here are correct, and if I enlarge the Kpoints choice in TranSIESTA, it only gives the nearly same result. At this case, if I do TBtrans calculation with $N_k^{TS} \leq N_k^{tbt}$, it seems everything goes well and sounds reasonable.
-
It tells me that k point sampling in TBtrans should generally be much denser. However, with Monkhost-Pack method both in TranSIESTA and TBtrans, I think the only way to enlarge kpoint density is to enlarge the number in block kgrid.MonkhorstPack, and the consequent kpoint number would be denser but more. So it seems that it’s hard to realize $N_k^{TS} \geq N_k^{tbt}$ and k point sampling in TBtrans should generally be much denser at the same time?
Thank you !