Skip to Content.
Sympa Menu

forum - Re: [abinit-forum] elphon - Transport memory requirements

forum@abinit.org

Subject: The ABINIT Users Mailing List ( CLOSED )

List archive

Re: [abinit-forum] elphon - Transport memory requirements


Chronological Thread 
  • 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
>
>




Archive powered by MHonArc 2.6.15.

Top of Page