Page 1 of 1

hwo to continue a job in one dataset ?

Posted: Fri May 06, 2011 5:36 am
by mazuju
hello,
I have run a job about computation of non-linear optical susceptibilities for two months, unfortunately, the job was killed by my careless manager.
The job was stopped in DATASET4, and most job of this DATASET has been completed.
If I want to continue the job from the end of DATASET4 , how to do that ?
Thanks !


the input file is supplied :

Code: Select all

# Linear and nonlinear response calculation 
# Perturbations: electric fields & atomic displacements
# (M. Veithen, 21.4.2005)

#Definition of lattice parameters
#********************************
acell     2.1514196567E+01  1.0930762115E+01  2.1913430467E+01 Bohr
angdeg    90.00000000  89.15405976  90.00000000

#Definition of atomic positions and types
#*************************************
   natom       32
   ntypat      3
   znucl       11 33 34
   typat       8*1 8*2 16*3
   xred              9.7952872705E-01  4.7802694231E-01  7.6400207817E-01
                     2.3853805048E-01  9.8178542576E-01  2.4386528385E-01
                     4.7998601442E-01  5.0972762118E-01  7.3663927906E-01
                     7.3818964742E-01  5.2768042659E-01  5.0131769572E-01
                     9.7952872705E-01 -4.7802694231E-01  1.2640020782E+00
                     2.3853805048E-01 -9.8178542576E-01  7.4386528385E-01
                     4.7998601442E-01 -5.0972762118E-01  1.2366392791E+00
                     7.3818964742E-01 -5.2768042659E-01  1.0013176957E+00
                     7.4380025684E-01  9.3299475525E-01  7.7576256468E-01
                     2.9307365002E-04  9.1639727973E-01 -4.2041522417E-03
                     2.4094987151E-01  5.4607695553E-01  9.6440799078E-01
                     4.9815309108E-01  9.8226685013E-01  9.5348113633E-01
                     7.4380025684E-01 -9.3299475525E-01  1.2757625647E+00
                     2.9307365002E-04 -9.1639727973E-01  4.9579584776E-01
                     2.4094987151E-01 -5.4607695553E-01  1.4644079908E+00
                     4.9815309108E-01 -9.8226685013E-01  1.4534811363E+00
                     9.6396337901E-01  9.7936941322E-01  7.9685495820E-01
                     2.2929107707E-01  5.1868627655E-01  7.6630383948E-01
                     4.6092710204E-01  5.8373017861E-01  9.8874037255E-01
                     5.0203857945E-01  9.8897530069E-01  2.5480739275E-01
                     7.1231004160E-01  9.7372615953E-01  9.8154633776E-01
                     2.1614239417E-01  9.5830849506E-01  9.9433331845E-01
                     9.9806264760E-01  4.8087697596E-01  5.1414702901E-01
                     7.3031610201E-01  5.3770831975E-01  7.5162494626E-01
                     9.6396337901E-01 -9.7936941322E-01  1.2968549582E+00
                     2.2929107707E-01 -5.1868627655E-01  1.2663038395E+00
                     4.6092710204E-01 -5.8373017861E-01  1.4887403726E+00
                     5.0203857945E-01 -9.8897530069E-01  7.5480739275E-01
                     7.1231004160E-01 -9.7372615953E-01  1.4815463378E+00
                     2.1614239417E-01 -9.5830849506E-01  1.4943333184E+00
                     9.9806264760E-01 -4.8087697596E-01  1.0141470290E+00
                     7.3031610201E-01 -5.3770831975E-01  1.2516249463E+00
   ixc         3  # ???

#Parameters of the SCF cycles
#****************************
   iscf        7
   nstep       100

#Plane wave basis and k-point grid
#*********************************
   ecut        45
   ecutsm      0.5
   dilatmx     1.05
   ngkpt       4 6 4
   nshiftk     1
   shiftk      0.5 0.5 0.5


ndtset 5
#DATASET1 : scf calculation: GS WF in the BZ

#********************************************
   prtden1    1
   kptopt1    1
   toldfe1    1.0d-12

#DATASET2 : non scf calculation: GS WF in the whole BZ
#*****************************************************
   getden2    1
   kptopt2    2
     iscf2   -2
   getwfk2    1
   tolwfr2    1.0d-22
    nband2    72 # is it need to set nband in dataset1 ??


#DATASET3 : ddk on the MP grid of k-points
#*****************************************
   getwfk3    2
    rfdir3    1 1 1
   rfelfd3    2
   tolwfr3    1.0d-22
    nband3    72
   kptopt3    2

#DATASET4 : ddE on the MP grid of k-points
#*****************************************
   prtden4    1
   getwfk4    2
   getddk4    3
    rfdir4    1 1 1
   rfphon4    1
  rfatpol4    1 32
   tolvrs4    1.0d-12
   rfelfd4    3
    nband4    72
   kptopt4    2
  prepanl4    1

#DATASET5 : 3DTE calculation
#***************************
   getden5    1
  get1den5    4
   getwfk5    2
   get1wf5    4
    nband5    72
   kptopt5    2
optdriver5    5
  rf1elfd5    1
  rf1phon5    1
 rf1atpol5    1 32
   rf1dir5    1 1 1
  rf2elfd5    1
   rf2dir5    1 1 1
  rf3elfd5    1
   rf3dir5    1 1 1

nbdbuf 0

Re: hwo to continue a job in one dataset ?

Posted: Fri May 06, 2011 2:01 pm
by david.waroquiers
Hello,

You can restart from dataset 4 by setting ndtset 2 jdtset 4 5 (meaning, there are two datasets and their numbers are 4 and 5) but you cannot restart dataset 4 where it stopped.

David

Re: hwo to continue a job in one dataset ?

Posted: Fri May 06, 2011 2:11 pm
by mazuju
david.waroquiers wrote:Hello,

You can restart from dataset 4 by setting ndtset 2 jdtset 4 5 (meaning, there are two datasets and their numbers are 4 and 5) but you cannot restart dataset 4 where it stopped.

David


David,
Thanks for your advice. I know the way you said. I want to know how to restart dataset4 from where it stopped, because dataset 4 has spend 40 days.
Thank you all the same.

Re: hwo to continue a job in one dataset ?

Posted: Sat May 07, 2011 12:38 pm
by jzwanzig
If the job that stopped was doing response function calculations, you can just re-do the ones that didn't finish yet. You can figure this out (it's fairly tricky, though) by looking at the output files (they would have names like foo_DS4_1WFxx) and comparing them against the list of ipert in the output file for dataset 4 that abinit said it needed to do by symmetry). Then you could identify based on the abinit numbering scheme of perturbations (see the respfn help file) which ones remained to do, and run only those. What was your dataset 4 supposed to be calculating?