[Re:12] Regarding afm fcc-Fe convergance and higher total energy values compared to bcc fm-Fe

Posted on : December 10, 2018 (Mon) 20:46:21

by Hitoshi GOMI

Dear all,

KKR administrator send me the deleted log.

Regards,
Hitoshi GOMI


------------------------------------------------------------
Anti-Ferromagnetic (AFM) calculation for bcc and fcc-Fe
------------------------------------------------------------
Dear all,

I need your help to perform electronic structure calculation for bcc-Fe and fcc-Fe in anti-ferromagnetic state. I like to calculate the energy differences of anti-ferromagnetic and ferromagnetic bcc and fcc Fe.

My input files for bcc-Fe is following;

======================================================================
bcc-Fe
=======================================================================
c----------------------Fe------------------------------------
go data/fe
c------------------------------------------------------------
c brvtyp a c/a b/a alpha beta gamma
bcc 5.32 , , , , , ,
c------------------------------------------------------------
c edelt ewidth reltyp sdftyp magtyp record
0.001 1.0 nrl mjw mag init
c------------------------------------------------------------
c outtyp bzqlty maxitr pmix
update 4 100 0.035
c------------------------------------------------------------
c ntyp
2
c------------------------------------------------------------
c type ncmp rmt field mxl anclr conc
Fe 1 1 0.0 2
26 100
Fe2 1 1 0.0 2
26 100
c------------------------------------------------------------
c natm
2
c------------------------------------------------------------
c atmicx(in the unit of a) atmtyp
0a 0b 0c Fe
0.5a 0.5b 0.5c Fe2


============================================================

and for fcc-Fe is below;

========================================================================
fcc-Fe
=======================================================================
c----------------------Fe------------------------------------
go data/fe
c------------------------------------------------------------
c brvtyp a c/a b/a alpha beta gamma
fcc 5.32 , , , , , ,
c------------------------------------------------------------
c edelt ewidth reltyp sdftyp magtyp record
0.001 1.0 nrl mjw mag init
c------------------------------------------------------------
c outtyp bzqlty maxitr pmix
update 4 100 0.035
c------------------------------------------------------------
c ntyp
4
c------------------------------------------------------------
c type ncmp rmt field mxl anclr conc
Fe 1 1 0.0 2
26 100
Fe2 1 1 0.0 2
26 100
Fe3 1 1 0.0 2
26 100
Fe4 1 1 0.0 2
26 100
c--------------------------------------------------------
c natm
4
c------------------------------------------------------------
c atmicx(in the unit of a) atmtyp
0a 0b 0c Fe
0.5a 0.5b 0c Fe2
0.5a 0b 0.5c Fe3
0a 0.5b 0.5c Fe4
c------------------------------------------------------------

When the code starts to run then it does not put the atoms exactly to the position as specified by the input file. For example for bcc-Fe it shows following;

atoms in the unit cell
position= 0.00000000 0.00000000 0.00000000 type=Fe
position= 0.25000000 0.25000000 0.25000000 type=Fe2


and for fcc-Fe it shows following positions;

atoms in the unit cell
position= 0.00000000 0.00000000 0.00000000 type=Fe
position= 0.25000000 0.25000000 0.50000000 type=Fe2
position= 0.25000000 0.50000000 0.25000000 type=Fe3
position= 0.50000000 0.25000000 0.25000000 type=Fe4

Furthermore self-consistent iterations are also not converging for both cases and following warning also prompts to the screen during calculation;

***wrn in chklat...given rmt's conflict; reduced.

I therefore request to please suggest me correct method to perform calculation for anti-ferromagnetic state of bcc and fcc-Fe. I shall be thankful to you.





------------------------------------------------------------
[Re:01] Anti-Ferromagnetic (AFM) calculation for bcc and fcc-Fe
------------------------------------------------------------
Hello Qais Ali,

There are three problems to be solved.
(1) Atomic position
(2) Bravais lattice
(3) Initial potential

Let's consider the ferromagnetic bcc Fe and B2-like antiferromagnetic Fe.
Then, you can apply to the ferromagnetic fcc Fe and L10-like antiferromagnetic Fe.

(1) Atomic position
First, you have to understand the difference between "0.5 0.5 0.5 Fe2" and "0.5a 0.5b 0.5c Fe2" for the atomic positions.
The former is specified by the Cartesian coordinates, and the latter is the fractional coordinates.


(2) Bravais lattice
The crystal structure is specified by the combination of a Bravais lattice and a basis.
You set the bcc Bravais lattice; its primitive translation vectors are found in the output as
a=( -0.50000 0.50000 0.50000)
b=( 0.50000 -0.50000 0.50000)
c=( 0.50000 0.50000 -0.50000)

And you specified the basis with two Fe atoms, which is located at (0a 0b 0c) and (1/2a 1/2b 1/2c) in the fractional coordinates.
Therefore, the atomic position of Fe2 is 0.5a + 0.5b + 0.5c = (0.25 0.25 0.25) in the Cartesian coordinates.

I recomend you to use the sc (simple cubic) Bravais lattice and the Cartesian coordinates as follows.

c----------------------Fe------------------------------------
go data/B2AFMfe
c------------------------------------------------------------
c brvtyp a c/a b/a alpha beta gamma
sc 5.32 , , , , , ,
c------------------------------------------------------------
c edelt ewidth reltyp sdftyp magtyp record
0.001 1.0 nrl mjw mag 2nd
c------------------------------------------------------------
c outtyp bzqlty maxitr pmix
update 4 100 0.035
c------------------------------------------------------------
c ntyp
2
c------------------------------------------------------------
c type ncmp rmt field mxl anclr conc
Fe1 1 1 0.0 2
26 100
Fe2 1 1 0.0 2
26 100
c------------------------------------------------------------
c natm
2
c------------------------------------------------------------
c atmicx(in the unit of a) atmtyp
0 0 0 Fe1
0.5 0.5 0.5 Fe2
c------------------------------------------------------------

(3) Initial potential
By using the above input, you can calculate B2 Fe.
But it may becomes ferromagnetic.
To get the initial potential for anti-ferromagnetic Fe, you can refer the following thereads.
http://kkr.issp.u-tokyo.ac.jp/bbs/thread.php?id=258
http://kkr.issp.u-tokyo.ac.jp/bbs/thread.php?id=215
http://kkr.issp.u-tokyo.ac.jp/bbs/thread.php?id=115

Hitoshi GOMI




------------------------------------------------------------
[Re:02] Anti-Ferromagnetic (AFM) calculation for bcc and fcc-Fe
------------------------------------------------------------
Dear Hitoshi,

Thanks a lot for the very clear and quick response.

Now, I am able to run the 2 atoms bcc FM Fe with your suggested input file.

I also used fmg to create new potential that has AFM configuration as follows:

./data/Fe_fm 1 2
./data/Fe_afm 1 -2

Later, I changed my input file as like in below:

c---------------------Fe------------------------------------
go data/Fe_afm
c------------------------------------------------------------
c brvtyp a c/a b/a alpha beta gamma
sc 5.32 , , , , , ,
c------------------------------------------------------------
c edelt ewidth reltyp sdftyp magtyp record
0.001 1.0 nrl mjw mag 2nd
c------------------------------------------------------------
c outtyp bzqlty maxitr pmix
update 4 100 0.035
c------------------------------------------------------------
c ntyp
2
c------------------------------------------------------------
c type ncmp rmt field mxl anclr conc
Fe1 1 1 0.0 2
26 100
Fe2 1 1 0.0 2
26 100
c------------------------------------------------------------
c natm
2
c------------------------------------------------------------
c atmicx(in the unit of a) atmtyp
0 0 0 Fe1
0.5 0.5 0.5 Fe2
c------------------------------------------------------------


However, even if I start with AFM configured potential my calculations converge to 2 atoms down FM configuration.

Can you please give further feedbacks.

Thanks in advance,
Qais




------------------------------------------------------------
[Re:03] Anti-Ferromagnetic (AFM) calculation for bcc and fcc-Fe
------------------------------------------------------------
Hello Qais Ali,

Generally speaking, it is difficult to get the convergence of unstable state.
In this particular case, the fixed spin moment (fsm) mode may solve the problem.

Please try the following input with afm initial potential.
The last "0.0" is the bulk magnetic moment, which restrict the system non-magnetic.

Hitoshi GOMI

c---------------------Fe------------------------------------
fsm data/Fe_afm
c------------------------------------------------------------
c brvtyp a c/a b/a alpha beta gamma
sc 5.32 , , , , , ,
c------------------------------------------------------------
c edelt ewidth reltyp sdftyp magtyp record
0.001 1.0 nrl mjw mag 2nd
c------------------------------------------------------------
c outtyp bzqlty maxitr pmix
update 4 100 0.035
c------------------------------------------------------------
c ntyp
2
c------------------------------------------------------------
c type ncmp rmt field mxl anclr conc
Fe1 1 1 0.0 2
26 100
Fe2 1 1 0.0 2
26 100
c------------------------------------------------------------
c natm
2
c------------------------------------------------------------
c atmicx(in the unit of a) atmtyp
0 0 0 Fe1
0.5 0.5 0.5 Fe2
c------------------------------------------------------------
c moment
0.0
c------------------------------------------------------------