You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Here is the code used for this calculation (based on the example of the co-optimization in Wannier.jl).
# ## Preparation
# Load the packages
using WannierIO
using Wannier
#using WannierPlots
#
win = read_win("up/cri3_up.win")
model_up = read_w90("up/cri3_up")
model_dn = read_w90("dn/cri3_dn")
#The spin-up and down overlap matrices is written in the same format as `amn`
#
Mud = read_amn("cri3_updn.mud");
# then assemble into a [`MagModel`](@ref)
model = Wannier.MagModel(model_up, model_dn, Mud)
#=
Now let's disentangle with spin overlap constraint.
Here `λs` is the Lagrange multiplier for the constraint.
=#
λs = 100.0
U_up, U_dn = disentangle(model, λs;max_iter=4000);
#=
The resulting spin-up and spin-down WFs have very similar centers and spreads,
however, their centers drift from the original positions which were centered
on atoms.
=#
omega(model, U_up, U_dn, λs)
# position operator
pos_up = Wannier.TBPosition(model_up, gauges=U_up)
pos_dn = Wannier.TBPosition(model_dn, gauges=U_dn)
# Hamiltonian
ham_up = Wannier.TBHamiltonian(model_up, U_up)
ham_dn = Wannier.TBHamiltonian(model_dn, U_dn)
# write to tb files
Wannier.write_w90_tb("up/cri3_up", ham_up, pos_up)
Wannier.write_w90_tb("dn/cri3_dn", ham_dn, pos_dn)
In the end, the Wannier spread and the overlap are printed.
The spreads and the overlap does not seem to be very consistent from first sight. For example, the spread of the 1st spin up/down Wannier functions are 4.258 and 4.922 , and the overlap is 0.9946. The difference between the Wannier spread doesn't look so small, and the overlap looks quite close to 1. Could you help me to understand this?
Another question, in the overlap section
WF <↑|↓>/Ų
Is the unit Ų or is it normalized to 1?
Thanks a lot!
Best regards,
HeXu
The text was updated successfully, but these errors were encountered:
Hello Junfeng,
I have some questions about the Wannier spreads and the overlap of the spin up/down Wannier functions when doing the co-optimization.
The files related to this quesion can be found at:
https://github.com/mailhexu/Wannier.jl-TB2J-tests/tree/main/data/CrI3/cowf
Here is the code used for this calculation (based on the example of the co-optimization in Wannier.jl).
In the end, the Wannier spread and the overlap are printed.
The spreads and the overlap does not seem to be very consistent from first sight. For example, the spread of the 1st spin up/down Wannier functions are 4.258 and 4.922 , and the overlap is 0.9946. The difference between the Wannier spread doesn't look so small, and the overlap looks quite close to 1. Could you help me to understand this?
Another question, in the overlap section
Is the unit Ų or is it normalized to 1?
Thanks a lot!
Best regards,
HeXu
The text was updated successfully, but these errors were encountered: