Loop.s: Difference between revisions

From Wiki
Jump to navigation Jump to search
Dkarhanek (talk | contribs)
mNo edit summary
Dkarhanek (talk | contribs)
No edit summary
 
(4 intermediate revisions by the same user not shown)
Line 1: Line 1:
This is an example loop for finding the lattice constant of an FCC metal (e.g. copper). The loop creates a series of <TT>POSCAR</TT> files with primitive cell size length specified in the <TT>seq</TT> command (3.1 to 4.0 &#197; with steps of 0.1 &#197;, i.e. 10 runs).
go back to [[Main Page]], [[Computational Resources]], [[Scripts]], [[Scripts for VASP]]
 
This is an example loop for finding the lattice constant of an FCC primitive cell (a rhomboeder: see the figure below for elements appearing in FCC lattice).  
 
[[Image:CrystalStructures.png|frame|none|alt=Crystal Structures|Crystal Structures of The Elements (for usual form at room temeperature).]]
 
The loop creates a series of <TT>POSCAR</TT> files with primitive cell size length specified in the <TT>seq</TT> command (3.1 to 4.0 &#197; with steps of 0.1 &#197;, i.e. 10 runs).


This cell size parameter, cell volume from <TT>OUTCAR</TT> and energy form <TT>OSZICAR</TT> are extracted into the output files.
This cell size parameter, cell volume from <TT>OUTCAR</TT> and energy form <TT>OSZICAR</TT> are extracted into the output files.


Evaluation using a proper EOS method (preferentially ''Birch-Murnaghan equation'') should follow.
Evaluation using a proper EOS method (preferentially the ''[http://en.wikipedia.org/wiki/Murnaghan_Equation Murnaghan equation]'' or ''Birch-Murnaghan equation'') should follow.
(Do not use quadratic regression! - it's not an equation of state (EOS)!!)
 
'''Important'''
Do not use quadratic regression (second order polynom)! - it's not an equation of state (EOS)!! Some tools for the non-linear fitting of Murnaghan Equation are listed online here: [http://courses.cit.cornell.edu/das248/equation_of_state.html Online Resources for Calculating the Equation of State])
 
'''Note'''
In the <TT>INCAR</TT> file, please take care that you set up a rigid calculation - <TT>NSW=1</TT> - and that the electronic density is properly converged - <TT>NELMIN=10</TT>. The <TT>KPOINTS</TT> should define a very dense ''k''-point grid (such as 10x10x10 or more).
 
The script <TT>loop.s</TT> itself:


  #!/bin/bash
  #!/bin/bash
Line 32: Line 45:
     0.000000000000000  0.000000000000000  0.000000000000000  T  T  T
     0.000000000000000  0.000000000000000  0.000000000000000  T  T  T
  !
  !
 
  echo "a= $L"
  echo "a= $L"
  mpirun -np $NSLOTS vasp
  mpirun -np $NSLOTS vasp
Line 46: Line 59:
  done
  done
   
   
cat SUMMARY.log
  rm -f CHG* WAVE*
  rm -f CHG* WAVE*


'''Note'''
--[[User:Dkarhanek|Dkarhanek]] 13:53, 15 December 2010 (CET)
In the <TT>INCAR</TT> file, please take care that you set up a rigid calculation - <TT>NSW=1</TT> - and that the electronic density is properly converged - <TT>NELMIN=10</TT>. The <TT>KPOINTS</TT> should define a very dense ''k''-point grid (such as 10x10x10 or more).

Latest revision as of 13:53, 15 December 2010

go back to Main Page, Computational Resources, Scripts, Scripts for VASP

This is an example loop for finding the lattice constant of an FCC primitive cell (a rhomboeder: see the figure below for elements appearing in FCC lattice).

Crystal Structures
Crystal Structures of The Elements (for usual form at room temeperature).

The loop creates a series of POSCAR files with primitive cell size length specified in the seq command (3.1 to 4.0 Å with steps of 0.1 Å, i.e. 10 runs).

This cell size parameter, cell volume from OUTCAR and energy form OSZICAR are extracted into the output files.

Evaluation using a proper EOS method (preferentially the Murnaghan equation or Birch-Murnaghan equation) should follow.

Important Do not use quadratic regression (second order polynom)! - it's not an equation of state (EOS)!! Some tools for the non-linear fitting of Murnaghan Equation are listed online here: Online Resources for Calculating the Equation of State)

Note In the INCAR file, please take care that you set up a rigid calculation - NSW=1 - and that the electronic density is properly converged - NELMIN=10. The KPOINTS should define a very dense k-point grid (such as 10x10x10 or more).

The script loop.s itself:

#!/bin/bash
#$ -N Cu-lattice
#$ -pe c8m24_ompi 8
#$ -cwd
#$ -o vasp-ompi.out
#$ -e vasp-ompi.err
#$ -S /bin/bash
 
. /etc/profile.d/modules.sh

module load vasp/5.2
export OMP_NUM_THREADS=1

for L in `seq -f %3.1f 3.2 0.1 4.0`
do
cat >POSCAR <<!
Copper bulk - Cu1
  $L
    0.500000000000000   0.500000000000000   0.000000000000000
    0.000000000000000   0.500000000000000   0.500000000000000
    0.500000000000000   0.000000000000000   0.500000000000000
  1 
Selective dynamics
Direct
    0.000000000000000   0.000000000000000   0.000000000000000   T   T   T
!

echo "a= $L"
mpirun -np $NSLOTS vasp

V=`grep "volume " OUTCAR | tail -1 | awk '{print $5}'`
E=`tail -1 OSZICAR | awk '{print $5}'`

printf "%3.1f %6.2f %13.7f\n" $L $V $E >> SUMMARY-LVE.log
printf "%3.1f %13.7f\n" $L $E >> SUMMARY-LE.log
printf "%6.2f %13.7f\n" $V $E >> SUMMARY-VE.log
cp OSZICAR OSZICAR-$L
cp  OUTCAR  OUTCAR-$L
done

rm -f CHG* WAVE*

--Dkarhanek 13:53, 15 December 2010 (CET)