Inconsistent wannier result between EPW-wannier and standalone wannier90

[tyj]

Dear developer,
I'm using EPW 5.9 and QE 7.4.1. I found that the wannier result from EPW wannier90 library mode and just wannier90.x after QE are different. The spread is much larger in EPW wannier with same inputs. I've did lots of tests, and the only way i can get same spread and good fitting is using all the bands. But this makes calculation at much slower. The time to finish the job is at least doubled. All other parameter tuning like num_iter, dis_num_iter etc doesn't work. The .nnkp file from QE and EPW-wannier are same, except the bottom exclude band part when I exclude band in EPW. Exclude band or not in wannier90 doesn't affect the result, but affect a lot in EPW-wannier. I'm wondering why they are different and is there a way to solve this?

I'm attaching the input/output files below. Seems that I can only attach 3 at once. I would attach more in the following reply. But the key difference is below.

-------------wannier90 result-------------------
Sum of centres and spreads ( 44.279903, 76.695044,187.233577 ) 132.35961475

Spreads (Ang^2) Omega I = 108.382807209
================ Omega D = 0.240292891
Omega OD = 23.736501071
Final Spread (Ang^2) Omega Total = 132.359601171


---------EPW wannier result------------------
Sum of centres and spreads ( 44.280481, 76.698225,182.756900 ) 208.36738709

Spreads (Ang^2) Omega I = 151.828839446
================ Omega D = 1.123960105
Omega OD = 55.414577446
Final Spread (Ang^2) Omega Total = 208.367376997

Best,
Yujia

[tyj]

more input/output files here. The scf.in and nscf.in are same in both cases.