[Pw_forum] occupation

Matteo Cococcioni matteo at umn.edu
Thu Jul 30 17:57:47 CEST 2009


Dear Ali,

on the wiki page of quantum-espresso you should be able to find a new 
tutorial on the calculation of U. Please download it
and run the scripts in there following the instructions.



ali kazempour wrote:
>
> Dear all
>
> From Linear response calculation ,I get for TiO2 bulk U=5.73, But when 
> I double the unit cell in z direction and again compute U ,I get three 
> different value for Ti atoms.(U=5.63,5.46,5.70).
>

what are these three values? I assume they are what you get with 
supercells of three different sizes. is this correct? if it is the only 
strange thing is that the intermediate value is lower than the other 
two. otherwise you got pretty nice convergence of U which is what you 
want to achieve. Is there any particular reason (apart computational 
cost) why you only enlarge the cell in z direction?

> Why these values are not same . what is the main reason? Does it in 
> numerical variation errors?
> ANother question:
> When we perturb D level by alpha, it give rise to change in d 
> occupancy . Is it correct that if we impose negative alpha , the 
> occupation becomes less than initial state .?
>

no. positive alpha -- > decrease in total n. but you need to look at the 
right n....

> for Ti isolated atom for apha=0.1  tr[ns(na)] after first-iteration 
> equal to the 0.002 and at the end of calculation is 2.327 while we 
> know that for Ti,  tr[ns(na)] =2.000. Does it seems to be correct.?
>
>      Number of iteration with fixed ns =  0
>      Starting ns and Hubbard U :
>  enter write_ns
> U( 1) =  0.0000
> alpha( 1) =  0.1000
> atom  1   Tr[ns(na)]=   2.0000000
> atom  1  spin  1
> eigenvalues:  0.4000000 0.4000000 0.4000000 0.4000000 0.4000000
>  eigenvectors
>  1   1.0000000  0.0000000  0.0000000  0.0000000  0.0000000
>  2   0.0000000  1.0000000  0.0000000  0.0000000  0.0000000
>  3   0.0000000  0.0000000  1.0000000  0.0000000  0.0000000
>  4   0.0000000  0.0000000  0.0000000  1.0000000  0.0000000
>  5   0.0000000  0.0000000  0.0000000  0.0000000  1.0000000
>  occupations
>  0.400  0.000  0.000  0.000  0.000
>  0.000  0.400  0.000  0.000  0.000
>  0.000  0.000  0.400  0.000  0.000
>  0.000  0.000  0.000  0.400  0.000
>  0.000  0.000  0.000  0.000  0.400
> atom  1  spin  2
> eigenvalues:  0.0000000 0.0000000 0.0000000 0.0000000 0.0000000
>  eigenvectors
>  1   1.0000000  0.0000000  0.0000000  0.0000000  0.0000000
>  2   0.0000000  1.0000000  0.0000000  0.0000000  0.0000000
>  3   0.0000000  0.0000000  1.0000000  0.0000000  0.0000000
>  4   0.0000000  0.0000000  0.0000000  1.0000000  0.0000000
>  5   0.0000000  0.0000000  0.0000000  0.0000000  1.0000000
>  occupations
>  0.000  0.000  0.000  0.000  0.000
>  0.000  0.000  0.000  0.000  0.000
>  0.000  0.000  0.000  0.000  0.000
>  0.000  0.000  0.000  0.000  0.000
>  0.000  0.000  0.000  0.000  0.000
> nsum =   2.0000000
>  exit write_ns
>

this is not the first iteration. this is what the initial guess of the 
on-site occupations the code does to start the calculation


>  Atomic wfc used for LDA+U Projector are NOT orthogonalized
>      Starting wfc are    9 atomic +    1 random wfc
>
>      total cpu time spent up to now is     13.34 secs
>
>      per-process dynamical memory:   326.3 Mb
>
>      Self-consistent Calculation
>
>      iteration #  1     ecut=    45.00 Ry     beta=0.70
>      CG style diagonalization
>      c_bands:  3 eigenvalues not converged
>      c_bands:  2 eigenvalues not converged
>      ethr =  1.00E-12,  avg # of iterations = 40.4
>  enter write_ns
> U( 1) =  0.0000
> alpha( 1) =  0.1000
> atom  1   Tr[ns(na)]=   0.0021401
> atom  1  spin  1
> eigenvalues:  0.0000001 0.0000001 0.0000001 0.0009883 0.0009883
>  eigenvectors
>  1   0.0000000 -0.8803392 -0.4743427  0.0000000  0.0013927
>  2   0.0000000 -0.4743432  0.8803400  0.0000000 -0.0000040
>  3   0.0000000  0.0012242  0.0006642  0.0000000  0.9999990
>  4  -0.8886820  0.0000000  0.0000000 -0.4585240  0.0000000
>  5   0.4585240  0.0000000  0.0000000 -0.8886820  0.0000000
>  occupations
>  0.001  0.000  0.000  0.000  0.000
>  0.000  0.000  0.000  0.000  0.000
>  0.000  0.000  0.000  0.000  0.000
>  0.000  0.000  0.000  0.001  0.000
>  0.000  0.000  0.000  0.000  0.000
>

this is the first iteration! the problem is, in my opinion, that you 
didn't start this calculation from wfc and potential saved from the scf 
unperturbed run, but from scratch (that's why the code has to make the 
initial guess for the n).

regards,

Matteo
> ---------------------------------- and end of the file is
> :
>      End of self-consistent calculation
>  enter write_ns
> U( 1) =  0.0000
> alpha( 1) =  0.1000
> atom  1   Tr[ns(na)]=   2.3270531
> atom  1  spin  1
> eigenvalues:  0.1574294 0.1574294 0.1574294 0.9273825 0.9273825
>  eigenvectors
>  1   0.0000000 -0.4359030 -0.7986559  0.0000000  0.4148945
>  2   0.0000000  0.7269732 -0.0406891  0.0000000  0.6854592
>  3   0.0000000 -0.5305643  0.6004109  0.0000000  0.5983379
>  4  -0.9258933  0.0000000  0.0000000 -0.3777851  0.0000000
>  5   0.3777851  0.0000000  0.0000000 -0.9258933  0.0000000
>  occupations
>  0.927  0.000  0.000  0.000  0.000
>  0.000  0.157  0.000  0.000  0.000
>  0.000  0.000  0.157  0.000  0.000
>  0.000  0.000  0.000  0.927  0.000
>  0.000  0.000  0.000  0.000  0.157
>
>
> Ali Kazempour
> Physics department, Isfahan University of Technology
> 84156 Isfahan, Iran. Tel-1: +98 311 391 3733
> Fax: +98 311 391 2376 Tel-2: +98 311 391 2375
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum
>   


-- 
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Matteo Cococcioni
Department of Chemical Engineering and Materials Science,
University of Minnesota
421 Washington Av. SE
Minneapolis, MN 55455
Tel. +1 612 624 9056    Fax +1 612 626 7246
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 



More information about the Pw_forum mailing list