Large system with plane-wave dft


Just Got Here
Hello!
I’m trying to calculate graphene by pw dft on cluster. My purpose is to find out how large system can be calculated in it. There are 3 nodes with 4 12-cores processors, but meanwhile I’m working on single node. I’ve run 60 and 122-atoms graphene sheet, that took half and one minutes considerable in serial mode and faster in parallel. But I couldn’t get such result with 572 atoms. The calculation was run at 24 processes and has been lasted for second day already, but no result or error reports gives. In output file it continues writing the next message:
== Energy Calculation ==

          ====== Grassmann conjugate gradient iteration ======
     >>>  ITERATION STARTED AT Wed Jul 18 14:38:15 2012  <<<
    iter.           Energy         DeltaE       DeltaRho 
    ------------------------------------------------------
      10    0.1207690943E+06   -0.28759E+03    0.97155E+01
 Warning: Lagrange Multiplier tolerance too high:  5.68434188608080149E-014
         +Try using a smaller time step
         +Gram-Schmidt being performed, spin:                    1
 Warning: Lagrange Multiplier tolerance too high:  5.68434188608080149E-014
         +Try using a smaller time step
         +Gram-Schmidt being performed, spin:                    1
 Warning: Lagrange Multiplier tolerance too high:  1.13686837721616030E-013
         +Try using a smaller time step
         +Gram-Schmidt being performed, spin:                    1
…
etc.

And more there are strange massages above, which weren’t in case of 122 and 60 atoms:
          ------
          auto-z
          ------
  autoz: too many torsions 
     1 autoz failed with cvr_scaling = 1.2 changing to 1.3
  autoz: too many torsions 
     2 autoz failed with cvr_scaling = 1.3 changing to 1.4
  autoz: too many torsions 
     3 autoz failed with cvr_scaling = 1.4 changing to 1.5
  autoz: too many torsions 
     4 autoz failed with cvr_scaling = 1.5 changing to 1.6
  autoz: too many torsions 
     5 autoz failed with cvr_scaling = 1.6 changing to 1.7
      warning. autoz generated    7 bonds for atom    4
      warning. autoz generated    7 bonds for atom    6
      warning. autoz generated    7 bonds for atom    8
      warning. autoz generated    7 bonds for atom   10
      warning. autoz generated    7 bonds for atom   12
      warning. autoz generated    7 bonds for atom   14
      warning. autoz generated    7 bonds for atom   16
      warning. autoz generated    7 bonds for atom   18
      warning. autoz generated    7 bonds for atom   20
      warning. autoz generated    7 bonds for atom   22
      warning. autoz generated    7 bonds for atom   24
      warning. autoz generated    7 bonds for atom   26
      warning. autoz generated    7 bonds for atom   28
…


My input file is (atoms coordinates are cut off):
echo
start c42h18pw
memory global 2000 mb stack 1000 mb heap 1000 mb
geometry units angstroms
H	0.336018	0.904000	0.000001
C	2.459512	0.710000	0.000000
…
C	7.378536	0.710000	0.000000
C	36.892681	33.369999	0.000000
H	39.016174	33.175999	0.000000
end
pspw
  simulation_cell
    boundary_conditions aperiodic
    lattice_vectors
 20.0 0.0 0.0
 0.0 20.0 0.0
 0.0 0.0 10.0  
    ngrid 34 34 34
  end
  xc b3lyp
end
task pspw energy

Could you please to help me with my problem? Thank you in advance.

Gets Around
It looks like your unit cell is too small.


Forum >> NWChem's corner >> General Topics