[Pw_forum] Problem with determination of U using linear response

Matteo Cococcioni matteo at umn.edu
Wed May 27 14:04:50 CEST 2015


Dear Prasenjit,

I do not know the answer to your question (it probably depends on the
details of the DOS of your system) but the situation does not look tragic
to me.
I guess you can always expect a departure from linearity when the
perturbation grows bigger. typically I use values within -0.1 and 0.1.
Why the interacting response is more well behaved than the other? I am not
sure. maybe it's because the screenening lowers the effective magnitude of
the perturbation and so it's like you have been using a smaller alpha.

Hope this helps,

Matteo

On Wed, May 27, 2015 at 9:13 AM, Prasenjit Ghosh <prasenjit.jnc at gmail.com>
wrote:

> Dear all,
>
> I am using linear response to determine U for Fe. In my system, there are
> two types of Fe atom, one in which the Fe is in an O octahedron and the
> second in which it is in an O terahedron.
> When I am plotting the occupation number of the Fe d as a function of
> perturbation alpha (-0.2 .le. alpha .le 0.2), for the non interacting
> electron system, I am not getting a linear plot for the whole range of
> alpha while for the interacting one it remains linear.
> For example when Fe is in the O octahedra, the response does not remain
> linear for alpha is greater than 0.05 or less than -0.05. However, for the
> interacting case, it remains linear for the whole regime (please refer to
> the attached plot for Fe in O octahedra).
> Further for this case for alpha=0, the occupation for the interacting and
> non interacting ones are not exactly same. While for the non-interacting
> one the occupation is 6.60012, it is 6.60162 for the interacting one.
>
> Similarly for the tethedra, for alpha=-0.2, the non-interacting response
> does not remain linear while for the interacting one it remains linear.
>
> I was wondering whether one has faced similar situations and if so can
> provide me some hints as to what might be the probable reason for the same?
>
> Below is the input file I am using for the calculation:
>
> for alpha in -0.20 -0.15 -0.10 -0.05 0.00 0.05 0.10 0.15 0.20
>
> do
>
> mkdir /scratch/pghosh/octfe/alpha$alpha
>
> cd /scratch/pghosh/octfe/alpha$alpha
>
> ethr=`grep ethr /home/pghosh/Ca2Fe2O5/octfe/scf.out | tail -1 | awk
> '{print $3}'`
>
> rm -rf Ca2Fe2O5-noU-vcrel-AFM.*
>
> cp -r ../Ca2Fe2O5-noU-vcrel-AFM.* .
>
> cat > scf-octFe.$alpha.in << EOF
>  &control
>     calculation = 'scf',
>     prefix = 'Ca2Fe2O5-noU-vcrel-AFM',
>     outdir = '/scratch/pghosh/octfe/alpha$alpha'
>     pseudo_dir = '/home/pghosh/pseudo'
>  /
>  &system
>     ibrav =  0,
>     celldm(1)=10.25176385
>     nat = 36,
>     ntyp = 5,
>     ecutwfc = 45
>     ecutrho = 320
>     nspin=2
>     starting_magnetization(2)=-1.0
>     starting_magnetization(3)=1.0
>     starting_magnetization(4)=-1.0
>     occupations='smearing'
>     degauss=0.001
>     smearing='gaussian'
>         lda_plus_u = .true.,
>     U_projection_type = 'atomic',
>     Hubbard_U(2) = 1.d-10
>     Hubbard_U(3) = 1.d-10
>     Hubbard_U(4) = 1.d-10
>     Hubbard_U(5) = 1.d-10
>     Hubbard_alpha(4) = $alpha
>  /
>  &electrons
>     mixing_beta = 0.3
>      conv_thr = 1.0d-8,
>     startingwfc = 'file',
>     startingpot = 'file',
>     diago_thr_init = $ethr
>  /
>
> ATOMIC_SPECIES
>  Ca 87.62 Ca.pw91-nsp-van.UPF
>  Fe1 47.867 Fe.pw91-sp-van_ak.UPF
>  Fe2 47.867 Fe.pw91-sp-van_ak.UPF
>  Fe3 47.867 Fe.pw91-sp-van_ak.UPF
>  O 15.999 O.pw91-van_ak.UPF
> ATOMIC_POSITIONS (crystal)
> Ca       0.479439516   0.108738866   0.023376810
> Ca       0.979439516   0.391261134   0.476623190
> Ca       0.520560484   0.608738866   0.976623190
> Ca       0.020560484   0.891261134   0.523376810
> Ca       0.520560484   0.891261134   0.976623190
> Ca       0.020560484   0.608738866   0.523376810
> Ca       0.479439516   0.391261134   0.023376810
> Ca       0.979439516   0.108738866   0.476623190
> Fe3      0.000000000   0.000000000  -0.000000000
> Fe1      0.500000000   0.500000000   0.500000000
> Fe1      0.000000000   0.500000000  -0.000000000
> Fe1      0.500000000  -0.000000000   0.500000000
> Fe2      0.948517158   0.250000000   0.935170308
> Fe2      0.448517158   0.250000000   0.564829692
> Fe2      0.051482842   0.750000000   0.064829692
> Fe2      0.551482842   0.750000000   0.435170308
> O        0.262959232   0.985181682   0.237331409
> O        0.762959232   0.514818318   0.262668591
> O        0.737040768   0.485181682   0.762668591
> O        0.237040768   0.014818318   0.737331409
> O        0.737040768   0.014818318   0.762668591
> O        0.237040768   0.485181682   0.737331409
> O        0.262959232   0.514818318   0.237331409
> O        0.762959232   0.985181682   0.262668591
> O        0.027189198   0.140319152   0.069344138
> O        0.527189198   0.359680848   0.430655862
> O        0.972810802   0.640319152   0.930655862
> O        0.472810802   0.859680848   0.569344138
> O        0.972810802   0.859680848   0.930655862
> O        0.472810802   0.640319152   0.569344138
> O        0.027189198   0.359680848   0.069344138
> O        0.527189198   0.140319152   0.430655862
> O        0.586951586   0.250000000   0.888451999
> O        0.086951586   0.250000000   0.611548001
> O        0.413048414   0.750000000   0.111548001
> O        0.913048414   0.750000000   0.388451999
> K_POINTS (automatic)
> 9 3 9 1 1 1
> CELL_PARAMETERS (alat)
>    0.991167212   0.000000000   0.000000000
>    0.000000000   2.746908521   0.000000000
>    0.000000000   0.000000000   1.023657125
> EOF
>
>
>
> With regards,
>
> Prasenjit
> --
> PRASENJIT GHOSH,
> IISER Pune,
> Dr. Homi Bhabha Road, Pashan
> Pune, Maharashtra 411008, India
>
> Phone: +91 (20) 2590 8203
> Fax: +91 (20) 2589 9790
>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://pwscf.org/mailman/listinfo/pw_forum
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20150527/41d536c3/attachment.html>


More information about the users mailing list