[Pw_forum] Re hubbard U and stable structure
Naol Regassa
greatnaol at gmail.com
Mon May 18 14:48:28 CEST 2009
Dear GS
Thank you for responding to my question.
In my question I said "more effective" because I seen some changes on
the band structure of cubic ferromagnetic by adding U.But not as
effective ( it won't make totally an insulator) as the one I have seen
on the (experimentally found) stable structure i.e. hexagonal.
On 5/18/09, pw_forum-request at pwscf.org <pw_forum-request at pwscf.org> wrote:
> Send Pw_forum mailing list submissions to
> pw_forum at pwscf.org
>
> To subscribe or unsubscribe via the World Wide Web, visit
> http://www.democritos.it/mailman/listinfo/pw_forum
> or, via email, send a message with subject or body 'help' to
> pw_forum-request at pwscf.org
>
> You can reach the person managing the list at
> pw_forum-owner at pwscf.org
>
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of Pw_forum digest..."
>
>
> Today's Topics:
>
> 1. Re: Hubbard U and small degauss ... (Gabriele Sclauzero)
> 2. Electron-Phonon: loading dVscf file when ldisp=.TRUE.
> (Andrea Marini)
>
>
> ----------------------------------------------------------------------
>
> Message: 1
> Date: Mon, 18 May 2009 10:12:25 +0200
> From: Gabriele Sclauzero <sclauzer at sissa.it>
> Subject: Re: [Pw_forum] Hubbard U and small degauss ...
> To: PWSCF Forum <pw_forum at pwscf.org>
> Message-ID: <4A111869.30202 at sissa.it>
> Content-Type: text/plain; charset=us-ascii; format=flowed
>
> Dear Naol,
>
> Naol Regassa wrote:
>> Dear all
>>
>> I have two questions to ask
>>
>> a) For the last times I was calculating the effect of Hubbard U on the
>> system YMnO_3 using QESPRESSO 4.10. I got an insulator band structure
>> using U = 10.5eV though it is completely metal without U (GGA). When I
>
> This is plausible. Though I don't know the physics of your compound, I can
> tell you that
> for simpler transition metal oxides (like FeO) you find the same change
> (i.e. from metal
> to insulator) after introducing the Hubbard U. Please have a look at Matteo
> Cococcioni's
> PhD thesis (from http://www.sissa.it/cm/phd.php) and the related paper (PRB
> 2005).
>
>> got this result I have used small degauss = 0.006. My friend told me
>> to use degauss = 0.015.
>
> Have you asked him the reason? I don't see any...
>
> In principle you should test convergence of the total energy (and other
> quantities you're
> interested in) w.r.t all the parameters which introduce significant
> approximations to your
> result. As you decrease degauss you'll reach the correct result for your
> system at T=0
> (chosing k-point grids accordingly), so I don't see any reason to choose
> higher degauss
> other that saving computer time.
>
> In the specific case of DFT+U, you have to be carefull to converge the local
> occupations
> too. They are very sensitive to the value of ecutwfc, but also degauss and
> number k-points
> have to be checked.
>
> Please tell me how possible (correct) is
>> that? I checked the convergence for degauss but it is increasing near
>> 0.015.
>
> What is increasing?
>
>
>>
>> b) Is adding U value more effective for the most stable structure ?
>
> Sorry I cannot get this question... in which sense "more effective"?
>
>> For the unstable states like for my case cubic ferromagnetic phase I
>> can't see any change?
>
> What are looking at? Maybe your question is clear to someone who's studying
> systems like
> yours, so it's because of me that I cannot get the point... but if you could
> be more
> precise maybe you would have more change to get help.
>
>
> Regards,
>
>
> GS
>
>
>
>>
>> Here is the input for the 'scf'
>> &control
>> calculation = 'scf',
>> restart_mode = 'from_scratch',
>> prefix = 'ymo_hcp_ferro'
>> tstress = .true.,
>> tprnfor = .true.,
>> pseudo_dir = '/home/MSC608/Desktop/QESPRESSO_PP',
>> outdir='/home/MSC608/Thesis/Hexagonal/Ferro-Hex/temp'
>> /
>> &system
>> ibrav = 0,
>> celldm(1) = 6.7,
>> nat = 10,
>> ntyp = 4,
>> ecutwfc = 32,
>> ecutrho = 340.0
>> nspin = 2
>> starting_magnetization(2) = 0.7,
>> starting_magnetization(3) = 0.7,
>> occupations = 'smearing',
>> smearing = 'marzari-vanderbilt',
>> degauss = 0.005,
>> lda_plus_u=.true.,
>> Hubbard_U(2) = 10.5,
>> Hubbard_U(3) = 10.5,
>> /
>> &electrons
>> diagonalization = 'david'
>> startingwfc = 'atomic+random'
>> mixing_beta = 0.3,
>> conv_thr = 1.0d-10
>> /
>> ATOMIC_SPECIES
>> Y 88.91 Y.pbe-nsp-van.UPF
>> Mn1 54.94 Mn.pbe-sp-van_mit.UPF
>> Mn2 54.94 Mn.pbe-sp-van_mit.UPF
>> O 15.9994 O.pbe-rrkjus.UPF
>>
>> CELL_PARAMETERS (alat)
>> 0.985236441 -0.266531554 -0.000316189
>> -0.266366323 0.985859979 0.000349871
>> 0.000781869 -0.000890980 3.186836665
>>
>> ATOMIC_POSITIONS (crystal)
>> Y -0.082663260 -0.082649328 0.000174846
>> Y -0.082874577 -0.082880779 0.499813069
>> Mn1 0.205305857 0.560245090 0.249986248
>> Mn2 0.561457263 0.207812583 0.750012603
>> O -0.027464907 -0.056329933 0.249979343
>> O -0.058482388 -0.031289340 0.750035308
>> O 0.247951998 0.582250730 0.083800805
>> O 0.582338704 0.248188482 0.916046842
>> O 0.582135578 0.247948396 0.583964864
>> O 0.247742343 0.582041831 0.416186070
>> K_POINTS automatic
>> 3 3 2 1 1 1
>>
>> Thanks alot in advance
>> Tamene Regassa
>> AAU
>> Ethiopia
>> _______________________________________________
>> Pw_forum mailing list
>> Pw_forum at pwscf.org
>> http://www.democritos.it/mailman/listinfo/pw_forum
>>
>
> --
>
>
> o ------------------------------------------------ o
> | Gabriele Sclauzero, PhD Student |
> | c/o: SISSA & CNR-INFM Democritos, |
> | via Beirut 2-4, 34014 Trieste (Italy) |
> | email: sclauzer at sissa.it |
> | phone: +39 040 3787 511 |
> | skype: gurlonotturno |
> o ------------------------------------------------ o
>
>
> ------------------------------
>
> Message: 2
> Date: Mon, 18 May 2009 10:09:58 +0200 (CEST)
> From: Andrea Marini <Andrea.Marini at roma2.infn.it>
> Subject: [Pw_forum] Electron-Phonon: loading dVscf file when
> ldisp=.TRUE.
> To: Pw_forum at pwscf.org
> Message-ID:
> <alpine.LFD.2.00.0905180949290.25064 at anomaly.roma2.infn.it>
> Content-Type: text/plain; charset="us-ascii"
>
> Dear all,
>
> I am encountering a problem when calculating lambda constants
> (PH/elephon.f90). Till now I used to
>
> 1. calculate first phonons and dVscf file by using
>
> elph=.false.,
> trans=.true.,
> ldisp=.false.,
>
> in the ph.x input file.
>
> 2. re-read the dvscf file using a larger k-point grid and changing in the
> ph.x input file
>
> elph=.true.,
> trans=.false.,
> ldisp=.false.,
>
> This method works great. If I compare the result of point 2 against the
> calculation 1 using elph=.true. the lambda factors are identical.
>
> Now I am trying to follow points 1 and 2 when ldisp=.TRUE.. Here the
> results of the calculation 2, when the dVscf file is read from disk, are
> different the case when phonons and lambda's are calculated in the same
> run.
> More importantly I get that the symmetry of the lambda factors are not
> respected for some q-points when the dVscf file is read from disk:
>
> < ldisp=.true., trans=.true., elph=.true.
>> ldisp=.true., trans=.false., elph=.true.
>
> < q = ( 0.500000000 -0.500000000 0.500000000 )
> <
> < omega( 1) = 3.751393 [THz] = 125.133833 [cm-1]
> < omega( 2) = 3.751393 [THz] = 125.133833 [cm-1]
> < omega( 3) = 9.121952 [THz] = 304.277585 [cm-1]
> <
>
> < lambda( 1)= 2.6170 gamma= 1077.54 GHz
> < lambda( 2)= 2.6170 gamma= 1077.54 GHz
> < lambda( 3)= 0.1901 gamma= 462.75 GHz
> ---
>> lambda( 1)= 31.9820 gamma=13168.61 GHz
>> lambda( 2)= 23.5894 gamma= 9712.95 GHz
>> lambda( 3)= 19.7009 gamma=47963.51 GHz
>
>
> Maybe I am doing something wrong. I would really appreciate is someone
> could help me.
>
> I have attached to this mail a tar.gz file with all inputs to reproduce
> the above described error in the simple case of Al bulk. All the runs take
> less than 5 minutes. I have already created all the links so if you want
> to try, simple
>
>> cd Alluminum_TEST/
>> cd CALCULATE_DVSCF_LDISP_TRUE/
>> pw.x < gs.in > gs.out
>> ph.x < ph.in > ph.out
>> cd ../READ_DVSCF_LDISP_TRUE/
>> pw.x < gs.in > gs.out
>> ph.x < ph.in > ph.out
>> cd CALCULATE_DVSCF_SINGLE_Q/
>> pw.x < gs.in > gs.out
>> pw.x < nscf.in > nscf.out
>> ph.x < ph.in > ph.out
>> cd READ_DVSCF_SINGLE_Q
>> pw.x < gs.in > gs.out
>> pw.x < nscf.in > nscf.out
>> ph.x < ph.in > ph.out
>
> The diff between the files CALCULATE_DVSCF_LDISP_TRUE/ph.out and
> READ_DVSCF_LDISP_TRUE/ph.out is saved as DIFF_LDISP_TRUE.
>
> Cheers
>
> Andrea
>
> --------------------------------------------------------------------------
> Andrea MARINI
>
> Physics Department, University of Rome "Tor Vergata" (Italy)
> - phone: +39-0672594894 - fax: +39-062023507 -
>
> - andrea.marini at roma2.infn.it -- http://www.yambo-code.org/people/andrea -
> -------------- next part --------------
> A non-text attachment was scrubbed...
> Name: al_test.tar.gz
> Type: application/x-gzip
> Size: 23746 bytes
> Desc: Al bulk TEST
> Url :
> http://www.democritos.it/pipermail/pw_forum/attachments/20090518/24c418b2/attachment.bin
>
> ------------------------------
>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum
>
>
> End of Pw_forum Digest, Vol 23, Issue 43
> ****************************************
>
More information about the Pw_forum
mailing list