The ABINIT Users Mailing List ( CLOSED )

["BRUNEVAL Fabien 201177" <fabien.bruneval@cea.fr>]

Dear Thanusit,

Please don't consider the values of the gaps output by abinit. In fact, this 
functionality often doesn't work, especially for metals. You should rather 
calculate the gaps on your own, using the last column of the GW results.

If you don't mind, I have some comments about your GW calculation:
1) Make sure that you have a pseudopotential for Titane that leaves the 
semicore states 3s 3p as valence states. GW method is very sensitive to that 
issue.
2) It is usually a good strategy to keep ecutwfn = ecutsigx, since the cost 
of the calculation is driven by the maximum of these two values.

Enjoy the GW!

Fabien





-------- Message d'origine--------
De: thanusit@kku.ac.th [mailto:thanusit@kku.ac.th]
Date: sam. 2/9/2008 11:36
À: forum@abinit.org
Objet : [abinit-forum] A question about GW calculation of a metalic alloy 
Ga4P3Ti
 
Dear Abinit users

I tried to calculate the GW corrections(using plasmon pole model) for a 
metallic alloy Ga4P3Ti. In a convergence test of the GW corrections at Gamma, 
I got the results as shown below. Please note the E0 values. The calculation 
was done using Abinit 5.4.3., both Parallel and sequencial version. Results 
are the same.

-------------------------------------------
 Perturbative Calculation

 k =    0.000   0.000   0.000
 Band     E0 <VxcLDA>   SigX SigC(E0)      Z dSigC/dE  Sig(E)    E-E0       E
   36   0.145 -12.115  -8.696   0.484   0.893  -0.120  -8.628   3.487   3.632
   37   0.913  -9.582  -3.039  -2.806   0.925  -0.081  -6.125   3.457   4.370

 E^0_gap          0.000
 E^GW_gap         3.487
 DeltaE^GW_gap    3.487
-------------------------------------------

From the ground state band calculation, there is a clear separation between 
band 36 and 37. I wonder  why "E^0_gap" is reported "0.000", instead of 
0.768eV (equals to 0.913-0.145 eV).

I looked through the "occ" variables. They indicate that, for this metalic 
alloy, band 36 is half-filled while band 37 is empty. Does the above result 
have anything to do with the half-filled metalic character?

Regards, 

Thanusit Burinprakhon


##Input Files##

ndtset 15

# Calculation of the Self-Energy matrix elements (GW corrections)

optdriver    4        
getkss       1         
getscr       1        
nband        80      

##variable for convergence test
ecutwfn: 5.0         
ecutwfn+ 5.0          

ecutsigx 10.0       
nkptgw   1           
kptgw    0.0 0.0 0.0  
bdgw     36 37       

ecut  100            
acell  3*10.5056      
rprim  1.0  0.0 0.0   
       0.0  1.0 0.0   
       0.0  0.0 1.0 
ntypat 3         
znucl 22 15 31
natom 8
typat 1 2 2 2 3 3 3 3
xred
     0.0000000000E+00  0.0000000000E+00  0.0000000000E+00
     5.0000000000E-01  5.0000000000E-01  0.0000000000E+00
     5.0000000000E-01  0.0000000000E+00  5.0000000000E-01
     0.0000000000E+00  5.0000000000E-01  5.0000000000E-01
     2.5979480250E-01  2.5979480250E-01  2.5979480250E-01
     2.5979480250E-01  7.4020519750E-01  7.4020519750E-01
     7.4020519750E-01  2.5979480250E-01  7.4020519750E-01
     7.4020519750E-01  7.4020519750E-01  2.5979480250E-01
 
nstep 100 
diemac 12.0        
nsym 0        
symmorphi 0
ixc 7
iscf 5
prtvol 1

<<winmail.dat>>