Menu

ExternalModelParameters

KKozlov

Last modified 4 weeks ago

Parameters for External Model

Section xmmodel defines external model. No consistency checks are performed for parameters. The order of parameters should be as below.

parts

  • is the list of headers in par file with the number of parameters in each line

mask | nummask

  • mask=1;0;...;0; is an array of 0 and 1 of the length that is the number of parameters (not used).
  • nummask=31 specifies the number of 1 in the array.

parms | numparms

  • parms=2;3;1;...;5; are the values of the parameters as integers (not used).
  • numparms=40 number of parameters.

dparms | numdparms

  • dparms=0.1;0.2;...;0.9 are the values of parameters as doubles (in case of convert=gemstat in par file in the left-to-right up-to-bottom order).
  • numdparms=40 number of parameters.

tweak | numtweak | valnottweak

  • tweak=1;0;1;...;1; define if the value can be changed.
  • numtweak=16 number of parameters to tweak. 1 will be inserted in first numtweak places, all others will be set to 0.
  • valnottweak=1;34;15; is the list of indices to be set to 0, all others will be set to 1;

lbound | vallbound

  • lbound=0;0;...;0; is the lower bound for each parameter individually.
  • vallbound=0 is the lower bound for all parameters.

hbound | valhbound

  • hbound=0;0;...;0; is the upper bound for each parameter individually.
  • valhbound=0 is the upper bound for all parameters.

Another form of definition:

partnames=x;beta;MB;

  • names of groups of parameters

partsizes=60; 55; 1;

  • number of parameters in each group

parmvals=200; 5; 5;

  • initial values for each group

parmtypes=1; 0; 0;

  • type of parameters: 1 - integer values to be produced by rounding, 2 - integer values to be produced by sorting, 0 - real values; 4 - integer values to be produced by special rounding. Methods are described in [ConvertDoubleToInt]

maskvals=0; 0; 0;

  • values for masks

tweakvals=1; 1; 1;

  • define if the value can be changed.

dparmvals=200; 5.5; 5;

  • initial values for each group

lboundvals=0; -10; 0.01;

  • lower bound for each group

hboundvals=1924; 10; 9;

  • upper bound for each group

limitvals=1; 1; 1;

  • defines if parameters are limited

scalevals=1; 1; 1;

  • scale factor for aparameters

command

  • is run to get the score. The par file is simply appended to this string.

convert

Defines the parameter's file format. The following formats are supported:

  • convert=sdf for molecules.
  • convert=gemstat for gemstat program: one line in a file, starting with parameter name, parameters are delimited by space
  • convert=dgemstat two lines in a file, starting with parameter name, parameters are delimited by space
  • convert=subset for integer indices
  • convert=subsubset for subset of integer indices
  • convert=octave to use in Octave with source command
  • convert=r to use in R with source command
  • convert=ini ini format to use with Qt framework
  • if this option is not given parameters are printed to the command line

delimiters

  • are used to parse the output of the command

keys

  • defines the indices of the elements of the parsed output that we need

mapping

  • defines the reordering of keys

Example:

[default_model]
partnames=x;beta;MB;
partsizes=60; 55; 1;
parmvals=200; 5; 5; 
parmtypes=1; 0; 0;
maskvals=0; 0; 0;
tweakvals=1; 1; 1;
dparmvals=200; 5.5; 5; 
lboundvals=0; -10; 0.01;
hboundvals=1924; 10; 9; 
limitvals=1; 1; 1;
scalevals=1; 1; 1;
command=script 
delimiters=\n=
keys=0;1;2;
convert=ini
mapping=0;0;1;2;
type=command

Related

Wiki: ConvertDoubleToInt
Wiki: Home

Want the latest updates on software, tech news, and AI?
Get latest updates about software, tech news, and AI from SourceForge directly in your inbox once a month.