Top Page > Browsing

Question regarding Wannier calculations Date: 2017/11/16 13:43 Name: Hashmi   <arqumhashmi@gmail.com> I am interested in calculating the Berry curvature and AHC. Openmx optimize the wannier functions by itself and then i have to optimized it again by Wannier90 (because of .chk file used in wannier90). My question is optimizing the wannier by openmx can help to optimize and converge the MLWF on wannier90??? or i am just doing optimization two times and doing double work, first optimize it by openmx and then by wannier90. I am confused regarding this issue and i will really appreciate your reply.

Page: [1]

Re: Question regarding Wannier calculations ( No.1 ) Date: 2017/11/16 16:52 Name: Yo Pierre Mizuta Hi Hashmi, When you want to obtain all the Wannier-related quantities in Wannier90, you do not have to optimize Wannier functions within OpenMX. All you need to make OpenMX do is to generate *mmn, *amn, *eig files from initial projections. Please just specify two relevant keywords as below, and then run Wannier90. Wannier.Dis.SCF.Max.Steps 1 # default 200 Wannier.Minimizing.Max.Steps 1 # default 200 I hope this solves your question. Y.P. Mizuta Re: Question regarding Wannier calculations ( No.2 ) Date: 2017/11/17 11:20 Name: Hashmi  <arqumhashmi@gmail.com> Thanks miztua, I know it very well. But openmx can also optimize the wannier function, what is that for?? 2nd because Wannier90 can only run on serial so i wanna know if i will do optimizations of wannier firstly via openmx, can it help to optimize it on wannier90. If openmx wannier optimization can help on wannier90 then it's so far the best way to optimize wannier. Re: Question regarding Wannier calculations ( No.3 ) Date: 2017/11/23 12:27 Name: Yo Pierre Mizuta Dear Dr. Hashmi, I am sorry there was a lack of information I provided. "But openmx can also optimize the wannier function, what is that for?? " : --> Of course optimizing a set of WFs within OpenMX will give you the Hamiltonian represented with the basis set, which I think is an important blessing from this functionality. "because Wannier90 can only run on serial so i wanna know if i will do optimizations of wannier firstly via openmx, can it help to optimize it on wannier90." : --> Yes, but since the parallelization in Generate_Wannier() is focused on the initial part, where the overlap matrix is computed, I doubt whether it will make a big difference in computational time between using OpenMX or Wannier90 in the subsequent processes, i.e. for optimizing WFs. Nevertheless, it should be possible to complete the optimization in OpenMX, and directly move on to the calculation of Berry curvature & AHC using postw90. What we need for that is to generate either a binary file *.chk, or a text file *.chk.fmt to be converted to *.chk via w90chk2chk.x utility, exactly in the same format as the one Wannier90 does, so that postw90.x can read it. I have been trying to realize this by adding lines to Generate_Wannier.c, but not yet successful, due to something I am overlooking. I will let you know if I make progress on this line. Y.P. Mizuta Re: Question regarding Wannier calculations ( No.4 ) Date: 2017/11/27 10:59 Name: Hashmi  <arqumhashmi@gmail.com> Dear Mizuta yeah you got it exactly, what i wanted to know. you explained in detail and i understand it. Thanks again mizuta for your efforts. Yes, i am really interested and Please let me know if you solve this problem. Re: Question regarding Wannier calculations ( No.5 ) Date: 2017/12/14 10:28 Name: Chi-Cheng Lee  <cclee.physics@gmail.com> Hi Hashmi, If the Wannier functions can reproduce the bands you want, whether the Wannier functions are maximally localized or not will not change the physical quantities you want to study since we have gauge freedom for generating the Wannier functions, right? Why don't you just go ahead without optimizing them and write a good paper first? Cheers, Chi-Cheng

Page: [1]