[Wien] Error while running +/- 0.5e calculation

[Zhu, Jianxin]

Gavin,

I dug out my constraint LDA calculations for NiO done a few months ago.
The value of F_eff is 0.4293 Ry. I thought it is quite satisfactory in comparison with the value that Madsen and Novak published.
In my calculation, I used RKmax=5 with a single k-point.

Jianxin

From: Gavin Abo <gsabo at crimson.ua.edu<mailto:gsabo at crimson.ua.edu>>
Reply-To: A Mailing list for WIEN2k users <wien at zeus.theochem.tuwien.ac.at<mailto:wien at zeus.theochem.tuwien.ac.at>>
Date: Sunday, January 4, 2015 9:55 AM
To: A Mailing list for WIEN2k users <wien at zeus.theochem.tuwien.ac.at<mailto:wien at zeus.theochem.tuwien.ac.at>>
Subject: Re: [Wien] Error while running +/- 0.5e calculation

Probably, you cannot get the 'same' result (just something close to it).

Madsen and Novak got F_eff = 0.438 Ry [ http://www.wien2k.at/reg_user/textbooks/Constraint_U.pdf ].  However, keep in mind that they wrote this document in 2007, so the calculation was likely with a WIEN2k version 7 or 8 [ http://www.wien2k.at/reg_user/updates/ ].  Maybe, they were using 32 bit (probably more predominate at that time).  I expect they were using an older ifort/mkl (new at that time, but old today).   So differences in things like WIEN2k version, between 32 and 64 bit systems, and libraries (mkl, fftw) likely make it not possible to reproduce exactly.

Following the steps in Constraint_U.pdf, I got F_eff = ~0.4408 Ry.  When compared to Madsen and Novak's result, the percent error is less then 1 %, |(0.441-0.438)/0.438|*100% = 0.68 %).

The percent error in your case seems high (20 %):

F_eff = (4*-0.007129869+5*-0.027686778)/9-(4*(-0.332762453)+5*(-0.333786554))/9-0.43229+0.46699 = 0.349481024
|(0.349-0.438)/0.438|*100% = 20.3 %

As you can see in the table below, the big differences are coming from your :3DD001 and :3D 001 values for NiO_+.50.  So this suggests that the problem is with your NiO_+.50 calculation.



NiO_+.50

NiO_+.5-1



Mine

Yours

Mine

Yours

:FER

0.4779683788

0.43229

0.5182555839

0.46699

:3DD001

0.109619395

-0.007129869

-0.291937436

-0.332762453

:3D 001

0.126211972

-0.027686778

-0.273539538

-0.333786554


On 1/4/2015 4:41 AM, Abed Reg wrote:
Thank you Mr Gavin for your reply and especially for your interesting to my questions.

I have fixed the error and there was a problem in the structure file. Now every thing is ok but i didn't find the same results as in the paper of Madsen and novak.

My results are as follows:

1/ First result


--- FER -----------
in  1 files:
NiO_+.50.scf::FER  : F E R M I - ENERGY(GAUSS-.M.)=   0.43229
--- 3dd001 -----------
in  1 files:
NiO_+.50.scf::3DD001: 3D*                 -0.007129869 Ry



2/ Second result


--- FER -----------
in  1 files:
NiO_+.50.scf::FER  : F E R M I - ENERGY(GAUSS-.M.)=   0.43229
--- 3d -----------
in  1 files:
NiO_+.50.scf::3D 001: 3D                  -0.027686778 Ry



3/ Third result

--- FER -----------
in  1 files:
NiO_+.5-0.scf::FER  : F E R M I - ENERGY(GAUSS-.M.)=   0.46699
--- 3dd001 -----------
in  1 files:
NiO_+.5-0.scf::3DD001: 3D*                 -0.332762453 Ry



4/ Fourth result

--- FER -----------
in  1 files:
NiO_+.5-0.scf::FER  : F E R M I - ENERGY(GAUSS-.M.)=   0.46699
--- 3d -----------
in  1 files:
NiO_+.5-0.scf::3D 001: 3D                  -0.333786554 Ry



When i have applied the equation i did'nt find the same result as in the paper



My result is U= 0.314 Ry



Then i think i have made mistake but i don't know where?



My best regards


--


Mr: A.Reggad
Laboratoire de Génie Physique
Université Ibn Khaldoun - Tiaret
Algerie
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20150104/03242aad/attachment.html>