The ABINIT Users Mailing List ( CLOSED )

["Chun Li" <lichun@nuaa.edu.cn>]

Dear Masayoshi,

Thanks a lot for your reply. I had checked the *out file and *log file, and 
found that the alchemical mixing is performed by the code:
 ziontypat    6.00000000E+00  1.20000000E+01  6.50000000E+00
              1.15000000E+01
     znucl      8.00000   30.00000    8.00000    9.00000   29.00000   30.00000
But just like I mentionedd in the last email, it seemed that the code just 
considered the first 4 znucl in the calculation. That is, the typat=3 2 1 4 
made the the system as: O (-6)--Zn (+12)--O (-6)--F (-7), not the expected O 
(-6.5)--Zn (+12)--O (-6)--Zn (+11.5). In this meaning, the alchemical mixing 
had not been considered!

I tried changing the order of the atoms by setting znucl=8 30 9 8 29 30, then 
found the similar lines in the *out file:
 ziontypat    6.00000000E+00  1.20000000E+01  6.50000000E+00
              1.15000000E+01
     znucl      8.00000   30.00000    9.00000    8.00000   29.00000   30.00000
In this case, the typat=3 2 1 4 made the system as: F (-7)--Zn (+12)--O 
(-6)--O (-6), and this calculation was hard to converge of course. I think 
this test confirmed that the alchemical mixing was not considered indeed.

So I believe if it is not because of my missunderstanding of the variables 
settings for the alchemical method, then there must be some problem with the 
implementation of this method to ABINIT. Maybe someone with experiences like 
you can give me some ideas. Thank you very much!

Best regards.

Chun

----- Original Message ----- 
From: <mmikami@rc.m-kagaku.co.jp>
To: <forum@abinit.org>
Sent: Saturday, November 05, 2005 9:52 PM
Subject: [abinit-forum] Re: about alchemical mixing


> Dear Chun Li,
> 
> Regarding your post: 
> http://www.abinit.org/wws/arc/forum/2005-11/msg00006.html
> 
> I also notice the same bug in *_GEO files when alchemical computation is 
> perfomed. But you can confirm your compuation is really alchemically 
> performed by looking in your *.out/*.log file.
> For the time being, you can correct the computed *_GEO files manually with 
> the information in *.out/*.log. It should not be so difficult.
> 
> In passing, your alchemical mixing will NOT ALWAYS produce 
> physically-meaningful results, because your mixing is for heterovalent 
> mixing (O(2-)/F(1-), Cu(+)/Zn(2+))). You should confirm the result by 
> perfoming non-alchemical, standard computation (e.g. supercell with 
> assumption of Cu/F distribution). If your results could appear promising, 
> you would be welcome to come back here to let us know. ;-) I will like to 
> report such an example (not about your system) if/when it is accepted for 
> publication. (a part of my works is now submitted...) In my case, I had 
> performed the both computations, i.e., alchemical and non-alchemical one to 
> produce bands/DOS; they appear so close, incidentally (fortunately). I wish 
> I could disclose later (not for now) by registering the information at the 
> ABINIT bibliographical database page
> (http://www.abinit.org/community/?text=bib).
> 
> Regards,
> Masayoshi
> 
>