PAW on the Cobalt-Cluster


The Input File

General Remarks

The input file consists of keywords. They can be in any order and in free format.
The individual scripts read only that portion they are interested in. This means that every line
starting with a keyword that is not known is not read (which means that any misspelled keyword is
not read in, either!!!!)

Keywords ARE casesensitive and they are all lower case.
Two types of keyword must be distinguished:

oneline_keyw  <par1> <par2> ....

or

multiline_keyw
<.....
.....
....
...>
end

The only multiline keyword at the moment are coords and masses.
Most keywords have reasonable default settings supplied either from the scripts or the PAW program
itself. You always need coords and a directive.
If keywords are specified twice (by accident) the second setting usually overwrites the first.

The "new" scripts are basically derived from the "old" scripts (in ~program/paw/util/paw_scripts)  and
the old input files should still work with the "new" scripts. The major difference is the fact that
the keyword runtype is no longer necessary. The type of the run (dynamics or optimization) is
completely detemined by the directive. But since the runtype-keyword is simply overread now
it doesn't even have to be removed.
 

Keywords for make_cntl and make_strc

(This listing is currently not really ordered. I apologize)
parameters in square brackets can be left out [optional].
the first two keywords directive and coords must be present in any input file.

directive <switch>

switch =
for optimization:
conv_psi or conv_cont or conv_r
for linear transit calculations:
lt_move
for dynamics:
warmup or equil or sample

for optimizations:
the three options have to be used in that order (skip conv_cont if not present)
the "current" system is converged with the autopilot.
conv_psi:     start up. converge psi only
conv_cont:  start up continuum if present. converge psi and cont
conv_r:        start up nuclei motion. converge nuclei and psi (and cont if present)

for linear transit calculations:
lt_move:  with this directive basically an optimization run (conv_r) is performed with the following differences:


for dynamics:
warmup: switches the heatbath for the ions on and pulses exactly 150 steps to reach twice the
                   targettemp.
equil: will start up all thermostats to temp. the system runs steps timesteps.
sample: for a sample run an equil run must been done before. Neither the systems nor
               the thermostats are stoped. If there are values given in a constraint keyword the
                system tries to move  to that value in as many steps as given by the steps keyword.

coords <unit>
 <species> <x-coord> <y-coord> <z-coord> [optional: <solv-rad>]
 .
 .
end
the only multiline keyword
Specifies the coords (for conv_psi)
unit = angs, bohr  (default = angs)

species = the species name is matched in the paw.atomparams database.
A glance at the NEW (!!!) paw.atomparams reveals, that by e.g.
'rh' you get a default small core rh projector as you get it with 'rh17'. 'rh9' gives instead a
large core projector (with 9 valence electrons) which might be fine for optimizations.
IMPORTANT!!!! : Never use two different projectors for the same chemical element in the
same molecule. PAW can't handle this from the strc-file input structure and the script does not check this!

if solv-rad is not given, the dafault from paw.atomparams
is used (if continumm is on)
 

lattice <size> [optional: <type>]
specifies the size of the lattice (the actual size for fcc is sqrt(2)*size)
size = cell size in the unit specified in coords
type = fcc (no others at the moment, this is default)


charge <charge>

charge = charge of the system (default = 0)
currently no "unrestricted calculations are supported. The script produce
either a closed shell (even number of electrons) or a Doublett (Odd Number)


continuum <opt> <eps>

opt = on, off (default = off)
eps = epsilon for dielectricum (default =78.8)
dft_type <type>
type = lda, bp (default = bp)

lda means dft-type 1 and bp means 81 in paw (please
      check the paw-manual for that)


e_cutoff <cutoff>

 cutoff = cutoff value for psi in ryd. (default = 30)


steps   <steps>

 steps = number of steps for ONE paw run (default = 300)
 because of the fact that the autopilot stops the
 runs anyways, this value should be just set
 sufficiently large for optimizations.
 for a warmup run the number of steps is always 150.
auto_tol   <tolerance>
 tolerance = convergence criterion for the autopilot in ha (default = 0.00001)
 The autopilot stops when the total energy does not change more than the tolerance for
20 consecutive steps.
 
mass_scale  <scale>
scale = value to scale the chemical masses (from paw.atomaparams) for any dynamics run.
(default = 0.5)
For optimizations this value is not used.
masses
  <species>  <mass>
  .
  .
end
Multiline keyword!!!! (don't forget the end)
Masses to every species known (from the paw.atomparams database) can be assigned here.
This overrides any other setting in both dynamics and optimization runs.
Not all species from the coords must necessarily appear here and you can even specify masses for
other species which will have no effect, of course :-).
temp  <temperature>
temperature = temperature for the system in K (only used for dynamics runs).
For a warmup run the target temp. is set to 2*temp because of the energy dissipation into other
degrees of freedom. for equil and sample the thermostat will be set to this temperature.
(default = 300)
opt_fric  <friction>
friction = only for optimizations this is the friction of the other systems not controlled by the
                  autopilot. Should be small but larger than zero.
(default = 0.005)
For dynamics this value is not used.
timestep_opt  <step>
step = timestep for optimization runs in a.u.
(default = 10.0) (also determines the fictitous mass of psi)
For dynamics this value is not used.
timestep_dyn  <step>
step = timestep for dynamics runs in a.u.
(default = 7.0) (also determines the fictitous mass of psi)
For optimization this value is not used.
storepsir  <opt>
opt = on, off (default = on)
This switch determines whther the real space representation will be stored in memory (on) or
calculated on the fly (off). "off" saves some memory but slows the calcualtion down a bit.
c_substitution <atom1> <atom2> <atom3> [<value>]
atom1/2/3 = integer numbers referencing the atoms in the order they appear in the coords
                        section.
Important!!! this is not necesarily the order you see the atoms when you look at a movie
generated with prot2xyz, because PAW reorders the atoms.
This constraint maintains r(1->2) - r(2->3).
value = value of the constraint which is moved to (!!!) in case of a sample/lt_move run in steps timesteps.
c_bond <atom1> <atom2> [<value>]
atom1/2 = integer numbers referencing the atoms in the order they appear in the coords
                        section.
Important!!! this is not necesarily the order you see the atoms when you look at a movie
generated with prot2xyz, because PAW reorders the atoms.
This constraint maintains r(1->2).
value = value of the constraint which is moved to (!!!) in case of a sample/lt_move run in steps timesteps.
c_angle <atom1> <atom2> <atom3> [<value>]
atom1/2/3 = integer numbers referencing the atoms in the order they appear in the coords
                        section.
Important!!! this is not necesarily the order you see the atoms when you look at a movie
generated with prot2xyz, because PAW reorders the atoms.
This constraint maintains the angle a(1->2->3).
value = value of the constraint which is moved to (!!!) in case of a sample/lt_move run in steps timesteps.
c_torsion <atom1> <atom2> <atom3> <atom4> [<value>]
atom1/2/3/4 = integer numbers referencing the atoms in the order they appear in the coords
                        section.
Important!!! this is not necesarily the order you see the atoms when you look at a movie
generated with prot2xyz, because PAW reorders the atoms.
This constraint maintains the torsion angle (1->2->3->4).
value = value of the constraint which is moved to (!!!) in case of a sample/lt_move run in steps timesteps.
 
c_rotation <opt>
opt = on, off (default = off)
rotation constraint for the whole system
(translation constraint is always on)
 

Keywords for paw_run

There are some additional keywords which can modify the behaviour of the paw_run script.

restart_file    <opt>
report_file     <opt>
trajectory_file <opt>

If these keywords are found and opt = keep then the corresponding files will be copied back
to the homedirectory:
restart_file   ->  .rstrt copied as .restrt.new
report_file    ->   _constraint.report is copied back
trajectory_file ->  all files with the extension .tra are copied back

Rochus Schmid    (NEWS: 16.12.98)