Ironing out some Errors

Well, if you blindly followed my instructions correctly so far, you will probably run into some errors. So did I. It may be best to correct them one at a time, just so you can see examples of error output from the program should you run into these in the future. So, here are some errors that you may come across while trying to get it running for the first time, and suggested corrections to make:

• Wrong block structure:

  BLOCK STRUCTURE INCORRECT
  STOP IN BUFFER__READ
  BLOCK STRUCTURE INCORRECT
  FILE IDENTIFIER NOT RECOGNIZED
  VARIABLE ID HAS THE VALUE PROT
  STOP IN FILEHANDLER__UNIT
   EXIT: 2
  

  This is because I used the wrong structure for a block in the h2co.cntl file (forgot a !END. The input file parser recognized the error (which is in !PSIDYN !AUTO). Fix it by editing the file.

• Special characters in the input files:

  TABS AND OTHER SPECIAL LETTERS ARE NOT ALLOWED
  IN BUFFER STRUCTURED INPUT FILES
  VARIABLE LINE HAS THE VALUE !GENERIC    START=T NSTEP=100       !END
  VARIABLE POSITION HAS THE VALUE !GENERIC        START=T NSTEP=100       !END
  VARIABLE ASCII VALUE OF DISALLOWED CHARACTER HAS THE VALUE         9
  STOP IN BUFFER_READ
  TABS AND OTHER SPECIAL LETTERS ARE NOT ALLOWED
  IN BUFFER STRUCTURED INPUT FILES
  VARIABLE LINE HAS THE VALUE !GENERIC    START=T NSTEP=100       !END
  VARIABLE POSITION HAS THE VALUE !GENERIC        START=T NSTEP=100       !END
  VARIABLE ASCII VALUE OF DISALLOWED CHARACTER HAS THE VALUE         9
  FILE IDENTIFIER NOT RECOGNIZED
  VARIABLE ID HAS THE VALUE PROT
  STOP IN FILEHANDLER__UNIT
   EXIT: 2
  

  In both of my input files, I made it easier to read by using TABs to format the input, but this trips up the parser, too. The error message tells us the ascii value of the character that shouldn't be there, 9, which is a TAB. So, just replace all TABs with regular spaces, and save the file again.

• Orthogonality:

  LOOP FOR ORTHOGONALIZATION IS NOT CONVERGED
  STOP IN WAVES_ORTHO_X
   EXIT: 2
  

  This one is not so obvious. In general, it means that orthogonality cannot be maintained, probably because the waves are moving too fast. In this case, I forgot to limit the number of projectors and angular momentum in the center expansion. Otherwise, it tries to run with higher defaults to get higher accuracy. So, I need to set these limits.

  We have to edit the h2co.strc file to specify these limits under !SPECIES. After NAME=... we place npro=1 lrhox=1 for H, npro= 1 1 0 lrhox=1 for C, and npro= 1 1 lrhox=1 for O. See the manual p.52-53.

• Species masses:
  Since we are not worried about dynamics of the cores yet, we can make the masses of the specified atoms something that will give us better results for the optimization algorithm, rather than the true values. Set SPMquotM=2" (a.u.) for H and SPMquotM=5" for all other atoms in the .strc file. As a note, if you do not specify a mass, PAW will insert the correct one for you from its internal periodic table.

• Density Functional files:
  I also did not set the right files containing the partial waves, projector functions, etc, for each atom. If x is the proper species, what I really want to use is in

  $PAWDIR/Setups/PBE/00xx/x_.75_6.0.out

• Density Functional parameterization:
  Another thing to check is that the density functional parameterization that you tell PAW to use should match the method in the atomic files that you tell PAW to load. Otherwise, we get meaningless results. I neglected to include this parameter in my .strc file, so it went to the default (Perdew-Zunger parameterization of the Ceperley-Alder). Edit the file to make this Perdew-Burke-Ernzerhoff exchange and correlation:
  In the !CONTROL block, set !DFT type=10 !END. See manual p.18.

• Refer to the updated CNTL and STRC files in Figure 3 and Figure 4 respectively when you update your own input files.

  

Figure: h2co.cntl (2)

  

Figure: h2co.strc (2)