[Pw_forum] CaCuO2 and LDA+U

[Matteo Cococcioni]

Dear Chao Cao,

I'm not sure that I can provide very useful hints for this problem. with 
the few informations you provide it's a little bit like shooting in the 
dark....
First of all are you sure you are doing equivalent calculations with 
Vasp and PWscf? for example, are cutoffs the same? the smearing? what 
about potentials? Of course, qualitatively, results should not depend 
too much on the pseudopotential. But the story is a little bit different 
for LDA+U calculations. In some cases with different pseudos you may 
need quite different Us to obtain the same results (e.g. as large as 1 
eV differences may be needed switching from LDA to GGA). Did you try to 
"play" a little bit with U and check what happens for bigger values?
I checked your input file and it looks fine to me. However you use a 
curious unit cell that is rotated 45 degrees wrt the cartesian axes on 
the xy plane. Again this shouldn't matter in standard calculations. 
However with LDA+U sometimes the code gets trapped in local minima and 
the initial choice of occupations could be important. changing the 
orientation of the unit cell will change the orientation of the 
spherical harmonics used to obtain the atomic occupations. Maybe this 
will help finding the right ground state.
Apart from this Cu compounds may be problematic (especially with oxygen 
present). I have no direct experience yet but you can try to look for 
Paul Grant's email to the forum on the topic. How many electrons do you 
get on the Cu atoms?

Hope this helps.

Matteo



Chao Cao wrote:
> Dear PWSCF users and developers:
>
> I have been trying to use PWSCF's implementation of LDA+U to calculate 
> the electronic structure of CaCuO2 lately. In VASP, once I turn on U, it 
> gives AFM ground state and reasonable gap separation at E_F. However, 
> the procedure seems to be much harder for PWSCF. While it gets AFM gs 
> correctly, I can never get a band gap. I searched the forum, and I 
> realized that the initial occupation number might be important, but the 
> several combinations I tried seem to give me the same result. So, is 
> there any hint or suggestions, especially how to choose initial 
> occupation numbers? Thanks in advance.
>
>
> Best,
>
> Chao Cao
>
>
> And, here's my input file:
>
> &control
>   calculation = 'scf'
>   restart_mode = 'from_scratch'
>   prefix = 'CaCuO2'
>   disk_io = 'low'
>   outdir = './'
> /
> &system
>   ibrav = 0, celldm(1) = 1.88972688, ntyp = 4, nat = 16
>   nbnd=80, ecutwfc = 32, ecutrho = 400
>   nspin = 2, starting_magnetization(2)=0.5, starting_magnetization(3)=-0.5
>   lda_plus_u = .true., Hubbard_U(2) = 6.5, Hubbard_U(3) = 6.5
>   occupations = 'smearing', smearing = 'mp', degauss = 0.007
> /
> &electrons
>   mixing_beta = 0.2
>   conv_thr = 1.0d-9
> /
> ATOMIC_SPECIES
> Ca   1.0  Ca.pbe-nsp-van.UPF
> Cuu  1.0  Cu.pbe-n-van.UPF
> Cud  1.0  Cu.pbe-n-van.UPF
> O    1.0   O.pbe-van.UPF
> ATOMIC_POSITIONS crystal
> Ca   0.25  0.75  0.25
> Ca   0.75  0.25  0.25
> Ca   0.25  0.75  0.75
> Ca   0.75  0.25  0.75
> Cuu  0.25  0.25  0.00
> Cuu  0.75  0.75  0.50
> Cud  0.75  0.75  0.00
> Cud  0.25  0.25  0.50
> O    0.00  0.00  0.00
> O    0.50  0.00  0.00
> O    0.00  0.50  0.00
> O    0.50  0.50  0.00
> O    0.00  0.00  0.50
> O    0.50  0.00  0.50
> O    0.00  0.50  0.50
> O    0.50  0.50  0.50
> K_POINTS automatic
>   8,8,4,1,1,1
> CELL_PARAMETERS cubic
>    3.878393250   3.878393249   0.000000000
>   -3.878393249   3.878393250   0.000000000
>    0.000000000   0.000000000   6.409031058
>
>
>
>
>   

-------------- next part --------------
A non-text attachment was scrubbed...
Name: matteo.vcf
Type: text/x-vcard
Size: 294 bytes
Desc: not available
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20080325/c35fbcdd/attachment.vcf>