forum@abinit.org
Subject: The ABINIT Users Mailing List ( CLOSED )
List archive
- From: Marty Blaber <martin.blaber@gmail.com>
- To: forum@abinit.org
- Subject: Re: [abinit-forum] elphon - Transport memory requirements
- Date: Fri, 15 May 2009 09:53:40 +1000
- Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=mime-version:sender:in-reply-to:references:date :x-google-sender-auth:message-id:subject:from:to:content-type; b=bSiJPfD3Yi3iO+CsISAdrDZvrO8eh+dR3vTtlMFj5tkCDtcjjs6gxb4d1v61D4Z7xT ka4rZV0R06FOm6eo6I+u7H/siMDr+KdsLn1tWabRKlTAfP/rkZ0DiXmPZOnYSon2bMnU YzdUhovgVKmwO4zxCh6LRAjaXF+2ysEyEsk4M=
Hi Matthieu,
Sorry it took me so long to get back to you. Unfortunately there is a bug (typo) in the code. (Still present in 573)
It is only apparent when you turn on gkqwrite, probably left over from some previous testing.
On line 284 of 17ddb/normsq_gkq.F90
write (*,'(a,i4,2(3E16.6,3x))') 'gkk_qpt_tmp (band1,band1)', &
& iqptfull, gkk_qpt_tmp(:,1,:,1,1:10)
should be
write (*,'(a,i4,2(3E16.6,3x))') 'gkk_qpt_tmp (band1,band1)', &
& iqptfull, gkk_qpt_tmp(:,1,:,1,1:elph_ds%nsppol)
(I think)
Cheers,
Marty
gkk_qpt_tmp (band1,band1)
Segmentation Fault
Electron Phonon Coupling
Transport Spectral Function
electrical resistivity
thermal conductivity
2009/3/26 matthieu verstraete <matthieu.jean.verstraete@gmail.com>:
> 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
>
>
- Re: [abinit-forum] elphon - Transport memory requirements, Marty Blaber, 05/15/2009
- Re: [abinit-forum] elphon - Transport memory requirements, matthieu verstraete, 05/15/2009
Archive powered by MHonArc 2.6.15.