forum@abinit.org
Subject: The ABINIT Users Mailing List ( CLOSED )
List archive
- From: matthieu verstraete <matthieu.jean.verstraete@gmail.com>
- To: forum@abinit.org
- Subject: Re: [abinit-forum] elphon - Transport memory requirements
- Date: Thu, 26 Mar 2009 12:35:30 +0100
- Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :content-type:content-transfer-encoding; b=g1MriCJ9vriLp1WQomhbAWGsZU0zChbH/L1ZLvXv5y42aVn+ZF4oMp2CpAj2xfzvAG u2fFmfGKawRA5UW/+U8BTCqopM2LXZPbgoEQOdEh3i2Vwv758ul4spBgynkeVe/Dewx/ 2jHyxZ9dMJA0NK05IQzpJU/iQyF/thp1Q4i8g=
Hello,
nice to see people using this bit of the code. I'm afraid the
symmetries are not used enough for saving of memory etc.. (and there
are specific difficulties in doing so here). The answer is yes, you
need to keep all bands, because later the transport integrals contain
different Fermi speeds for different bands. All of the k and q are
kept in memory, and here all of the bands (2 band indices) as well.
However, you should try (with a later version of abinit, preferably)
to uncomment the prtgkq variable, which swaps much of the 100Gb to
disk (slower but ok as it is done in a structured way). This should
also work with the transport part.
keep us posted.
Matthieu
On Thu, Mar 26, 2009 at 5:43 AM, Marty Blaber <martin.blaber@gmail.com> wrote:
> Hi,
>
> I've been trying to perform an electron-phonon transport calculation
> on a 2 atom CsCl metallic system on a 16^3 kptrlatt with 35 qpoints.
> To perform the transport calculation we must set the tkeepbands to 1
> to keep all of the gkk bands in memory. Is this really necessary? My
> calculations need ~100GB of memory (As I have only 32 GB at the
> moment, the system pages awfully.)
> Is there a solution to this problem, my system isn't that big... am I
> missing something obvious/asking a stupid question?
> My anddb/abinit inputs are below.
>
> Any advice would be appreciated.
>
> Cheers,
> Marty
>
> p.s. Using abinip 5.6.4 for GS + Pert, tried anaddb 564 and 573
>
> ############################################################
> ANADDB.IN
> ############################################################
> # turn on calculation of the electron-phonon quantities
> elphflag 1
> ifltransport 1
> tkeepbands 1
> doscalprod 1
> #gkqwrite 1
>
> # Path in reciprocal space along which the phonon linewidths
> # and band structure will be calculated
> nqpath 1
> qpath
> 0.0 0.0 0.0
> 0.0 1/2 0.0
> 0.0 1 0.0
> 0.0 1/2 0.0
> 1/2 1/2 0.0
> 0.0 1/2 1/2
> 1/2 1/2 1/2
> 1/2 0.0 0.0
>
> # Irrelevant
> mustar 0.136
>
> # Minimalistic qpoint grid
> ngqpt 8 8 8
> nqshft 1
> q1shft 0.0 0.0 0.0
>
> nshiftk 1
> shiftk 0.0 0.0 0.0
> kptrlatt 16 0 0 0 16 0 0 0 16
> telphint 1
>
> # impose acoustic sum rule in a symmetric way
> asr 2
> dipdip 1
>
> # bravais lattice necessary
> brav 1
>
> #
> ifcflag 1
> ifcana 1
> # ifc for all atoms?
> natifc 0
> atifc 1 2 3
>
> # print dielectric matrix with freq dependence
> dieflag 0
>
> # print out eigenvectors and symmetrize dyn matrix
> eivec 1
>
> #Wavevector list number 1 (Reduced coordinates and normalization factor)
> nph1l 1
> qph1l
> 0.0000 0.0000 0.0000 1
>
> symdynmat 0
>
> ############################################################
> /ANADDB.IN
> ############################################################
>
>
> ############################################################
> ABINIT.IN
> ############################################################
> ###########################################################
> # Welcome to Marty Blaber's Abinit - EP Coupling input
> # file generator
> # Usage: python ab-build-ep-calc.py {qpoint file}
> ###########################################################
> # File = qp.list
> # nLines = 17
> # headerLines = 12
> # nqp = 5
>
> # File built on 2009-03-23 10:43:29.835767 by Martin G Blaber's
> abinit EP coupling input file generator
>
> #+**********************+
> #| DATASETS |
> #+**********************+
>
> # Number of datasets = 36
> #ndtset 35
> #jdtset 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
> 23 24 25 26 27 28 29 30 31 32 33 34 35 36
> #ndtset 19
> #jdtset 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36
> ndtset 6
> jdtset 31 32 33 34 35 36
>
> #DATASET 1 - Ground State
> iscf 3
> kptopt1 3
> tolwfr1 1.0d-18
> prtden1 1
> nline1 8
>
> #ds = 2 qp(0) =0.375 0.25 0.0
> iscf2 -3 tolwfr2 1.0d-14 getden2 1
> kptopt2 3 nqpt2 1 qpt2 0.375 0.25 0.0
> rfatpol3 1 1 rfdir3 1 0 0 rfphon3 1 tolvrs3 1.0e-5 getwfk3 1
> getwfq3 2 kptopt3 3 nqpt3 1 qpt3 0.375 0.25 0.0
> rfatpol4 1 1 rfdir4 0 1 0 rfphon4 1 tolvrs4 1.0e-5 getwfk4 1
> getwfq4 2 kptopt4 3 nqpt4 1 qpt4 0.375 0.25 0.0
> rfatpol5 1 1 rfdir5 0 0 1 rfphon5 1 tolvrs5 1.0e-5 getwfk5 1
> getwfq5 2 kptopt5 3 nqpt5 1 qpt5 0.375 0.25 0.0
> rfatpol6 2 2 rfdir6 1 0 0 rfphon6 1 tolvrs6 1.0e-5 getwfk6 1
> getwfq6 2 kptopt6 3 nqpt6 1 qpt6 0.375 0.25 0.0
> rfatpol7 2 2 rfdir7 0 1 0 rfphon7 1 tolvrs7 1.0e-5 getwfk7 1
> getwfq7 2 kptopt7 3 nqpt7 1 qpt7 0.375 0.25 0.0
> rfatpol8 2 2 rfdir8 0 0 1 rfphon8 1 tolvrs8 1.0e-5 getwfk8 1
> getwfq8 2 kptopt8 3 nqpt8 1 qpt8 0.375 0.25 0.0
>
> #ds = 9 qp(1) =0.5 0.25 0.0
> iscf9 -3 tolwfr9 1.0d-14 getden9 1
> kptopt9 3 nqpt9 1 qpt9 0.5 0.25 0.0
> rfatpol10 1 1 rfdir10 1 0 0 rfphon10 1 tolvrs10 1.0e-5 getwfk10
> 1 getwfq10 9 kptopt10 3 nqpt10 1 qpt10 0.5 0.25 0.0
> rfatpol11 1 1 rfdir11 0 1 0 rfphon11 1 tolvrs11 1.0e-5 getwfk11
> 1 getwfq11 9 kptopt11 3 nqpt11 1 qpt11 0.5 0.25 0.0
> rfatpol12 1 1 rfdir12 0 0 1 rfphon12 1 tolvrs12 1.0e-5 getwfk12
> 1 getwfq12 9 kptopt12 3 nqpt12 1 qpt12 0.5 0.25 0.0
> rfatpol13 2 2 rfdir13 1 0 0 rfphon13 1 tolvrs13 1.0e-5 getwfk13
> 1 getwfq13 9 kptopt13 3 nqpt13 1 qpt13 0.5 0.25 0.0
> rfatpol14 2 2 rfdir14 0 1 0 rfphon14 1 tolvrs14 1.0e-5 getwfk14
> 1 getwfq14 9 kptopt14 3 nqpt14 1 qpt14 0.5 0.25 0.0
> rfatpol15 2 2 rfdir15 0 0 1 rfphon15 1 tolvrs15 1.0e-5 getwfk15
> 1 getwfq15 9 kptopt15 3 nqpt15 1 qpt15 0.5 0.25 0.0
>
> #ds = 16 qp(2) =0.375 0.375 0.0
> iscf16 -3 tolwfr16 1.0d-14 getden16 1
> kptopt16 3 nqpt16 1 qpt16 0.375 0.375 0.0
> rfatpol17 1 1 rfdir17 1 0 0 rfphon17 1 tolvrs17 1.0e-5 getwfk17
> 1 getwfq17 16 kptopt17 3 nqpt17 1 qpt17 0.375 0.375 0.0
> rfatpol18 1 1 rfdir18 0 1 0 rfphon18 1 tolvrs18 1.0e-5 getwfk18
> 1 getwfq18 16 kptopt18 3 nqpt18 1 qpt18 0.375 0.375 0.0
> rfatpol19 1 1 rfdir19 0 0 1 rfphon19 1 tolvrs19 1.0e-5 getwfk19
> 1 getwfq19 16 kptopt19 3 nqpt19 1 qpt19 0.375 0.375 0.0
> rfatpol20 2 2 rfdir20 1 0 0 rfphon20 1 tolvrs20 1.0e-5 getwfk20
> 1 getwfq20 16 kptopt20 3 nqpt20 1 qpt20 0.375 0.375 0.0
> rfatpol21 2 2 rfdir21 0 1 0 rfphon21 1 tolvrs21 1.0e-5 getwfk21
> 1 getwfq21 16 kptopt21 3 nqpt21 1 qpt21 0.375 0.375 0.0
> rfatpol22 2 2 rfdir22 0 0 1 rfphon22 1 tolvrs22 1.0e-5 getwfk22
> 1 getwfq22 16 kptopt22 3 nqpt22 1 qpt22 0.375 0.375 0.0
>
> #ds = 23 qp(3) =0.5 0.375 0.0
> iscf23 -3 tolwfr23 1.0d-14 getden23 1
> kptopt23 3 nqpt23 1 qpt23 0.5 0.375 0.0
> rfatpol24 1 1 rfdir24 1 0 0 rfphon24 1 tolvrs24 1.0e-5 getwfk24
> 1 getwfq24 23 kptopt24 3 nqpt24 1 qpt24 0.5 0.375 0.0
> rfatpol25 1 1 rfdir25 0 1 0 rfphon25 1 tolvrs25 1.0e-5 getwfk25
> 1 getwfq25 23 kptopt25 3 nqpt25 1 qpt25 0.5 0.375 0.0
> rfatpol26 1 1 rfdir26 0 0 1 rfphon26 1 tolvrs26 1.0e-5 getwfk26
> 1 getwfq26 23 kptopt26 3 nqpt26 1 qpt26 0.5 0.375 0.0
> rfatpol27 2 2 rfdir27 1 0 0 rfphon27 1 tolvrs27 1.0e-5 getwfk27
> 1 getwfq27 23 kptopt27 3 nqpt27 1 qpt27 0.5 0.375 0.0
> rfatpol28 2 2 rfdir28 0 1 0 rfphon28 1 tolvrs28 1.0e-5 getwfk28
> 1 getwfq28 23 kptopt28 3 nqpt28 1 qpt28 0.5 0.375 0.0
> rfatpol29 2 2 rfdir29 0 0 1 rfphon29 1 tolvrs29 1.0e-5 getwfk29
> 1 getwfq29 23 kptopt29 3 nqpt29 1 qpt29 0.5 0.375 0.0
>
> #ds = 30 qp(4) =0.5 0.5 0.0
> iscf30 -3 tolwfr30 1.0d-14 getden30 1
> kptopt30 3 nqpt30 1 qpt30 0.5 0.5 0.0
> rfatpol31 1 1 rfdir31 1 0 0 rfphon31 1 tolvrs31 1.0e-5 getwfk31
> 1 getwfq31 30 kptopt31 3 nqpt31 1 qpt31 0.5 0.5 0.0
> rfatpol32 1 1 rfdir32 0 1 0 rfphon32 1 tolvrs32 1.0e-5 getwfk32
> 1 getwfq32 30 kptopt32 3 nqpt32 1 qpt32 0.5 0.5 0.0
> rfatpol33 1 1 rfdir33 0 0 1 rfphon33 1 tolvrs33 1.0e-5 getwfk33
> 1 getwfq33 30 kptopt33 3 nqpt33 1 qpt33 0.5 0.5 0.0
> rfatpol34 2 2 rfdir34 1 0 0 rfphon34 1 tolvrs34 1.0e-5 getwfk34
> 1 getwfq34 30 kptopt34 3 nqpt34 1 qpt34 0.5 0.5 0.0
> rfatpol35 2 2 rfdir35 0 1 0 rfphon35 1 tolvrs35 1.0e-5 getwfk35
> 1 getwfq35 30 kptopt35 3 nqpt35 1 qpt35 0.5 0.5 0.0
> rfatpol36 2 2 rfdir36 0 0 1 rfphon36 1 tolvrs36 1.0e-5 getwfk36
> 1 getwfq36 30 kptopt36 3 nqpt36 1 qpt36 0.5 0.5 0.0
>
> # Done
> #+----------------------+
> #| Constants |
> #+----------------------+
>
> #### Grid and Smearing ####
> occopt 4 # 4 = Metallic cold smear
> tsmear 0.001
> ecut 28.0
> nband 8
>
> #### K-POINTS ####
> #kptopt 3 # 3 = No Sym
> nshiftk 1
> shiftk 0.0 0.0 0.0
> kptrlatt 16 0 0 0 16 0 0 0 16
>
> #### SCF ####
> #iscf 3
> nstep 1000
> #toldfe 1.0d-8 # Energy
> #tolwfr 1e-18
>
>
> #### CELL ####
> acell 3*6.4474824015
> rprim 1.0 0.0 0.0 # SC primitive vectors
> 0.0 1.0 0.0
> 0.0 0.0 1.0
>
> #### ATOMS ####
> ixc 7 # 7 = PW92
> ntypat 2
> znucl 11 47
> natom 2
> typat 1 2
> xred 0.0 0.0 0.0
> 0.5 0.5 0.5
>
>
>
>
> ############################################################
> /ABINIT.IN
> ############################################################
>
--
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Dr. Matthieu Verstraete
European Theoretical Spectroscopy Facility (ETSF)
Dpto. Fisica de Materiales,
U. del Pais Vasco,
Centro Joxe Mari Korta, Av. de Tolosa, 72, Phone: +34-943018393
E-20018 Donostia-San Sebastian, Spain Fax : +34-943018390
Mail : matthieu.jean.verstraete@gmail.com
http://www-users.york.ac.uk/~mjv500
- [abinit-forum] elphon - Transport memory requirements, Marty Blaber, 03/26/2009
- Re: [abinit-forum] elphon - Transport memory requirements, matthieu verstraete, 03/26/2009
Archive powered by MHonArc 2.6.15.