Documentation of tbx3es1

Global Index (all files) (short | long) | Local contents | Local Index (files in subdir) (short | long)

Function Synopsis

GEAOPT = tbx3es1

Help text

 ToolBoX function to define parameters for locally oriented optimization of real variables

 This function defines parameters for the locally oriented 
 optimization of real valued variables inside the individuals.
 This proedure uses no recombination and mutation operators 
 from evolution strategies area.
 The locally oriented optimization only work with a small number
 of individuals which produce a lot of offspring. Only the best 
 of these are added to the population.
 In order to achieve this we have a choice of two possibilities.
 On the one hand, the population can contain few individuals 
 which each produce a lot of offspring. Only the best of these 
 offspring replace the parents and form the new population. 
 - population size: 1-5 individuals,
 - generation gap: 3-10 (number of offspring per parent),
 - selection pressure: 1 (no selection pressure).
 On the other hand, one works with a larger population size, 
 in which only the very best individuals produce offspring. 
 Here, (almost) all offspring replace the parents and form 
 the new population:
 - population size: 5-20 individuals,
 - generation gap: 1 (produce as many offspring as parents),
 - selection: truncation selection with a selection pressure 
   of 3-10 (only the very best individuals are selected and 
   each produce several offspring).
 The second variant has some small advantages, which have 
 particularly come to the fore during practical application.
 Thus, the second variant is used by default.
 However, the first variant is included in the comments below.

 Syntax:  GEAOPT = tbx3es1

 Input parameter:
    no input parameters

 Output parameter:
    GEAOPT    - Structure with newly defined options

 See also: muts1, mutes2, geamain2, geaoptset

Cross-Reference Information

Listing of function tbx3es1

% Author:   Hartmut Pohlheim
% History:  05.05.1999  file created
%           06.11.2005  two methods of selection options explained


function GEAOPT = tbx3es1


% Set function specific parameters
   GEAOPT = geaoptset(  'VariableFormat',       0           ...   % Use real valued variables 
                      , 'Recombination.Name',  {'recnone'}  ...   % no recombination used
                      , 'Mutation.Name',       {'mutes2'}   ...   % Define the evo. strategy real valued mutation function
                      , 'Mutation.Range',      [0.01]       ...   %  Initial mutation step size range
                     ...
                      , 'NumberSubpopulation',   1          ...   % Number of subpopulations (just one panmictic population)
                     ...
                      , 'NumberIndividuals',    12          ...   % Number of individuals per subpopulation
                      , 'Selection.Name',      {'seltrunc'} ...   % truncation selection must be used
                      , 'Selection.Pressure',        [6]    ...   % high selection pressure between parents (only best 1/6 parents are selected)
                      , 'Selection.GenerationGap',   [1]    ...   % Number of offspring per individual
                      , 'Selection.RankingMethod',    1     ...   % use non-linear ranking (necessary for the hihg selective pressure)
                     ... %  , 'NumberIndividuals',         [2]    ...  % Number of individuals per subpopulation
                     ... %  , 'Selection.Pressure',        [1]   ...   % no selection pressure between parents
                     ... %  , 'Selection.RankingMethod',    0     ...  % use linear ranking
                     ... %  , 'Selection.GenerationGap',   [6]   ...   % Number of offspring per individual
                     ...
                      , 'Selection.ReinsertionRate', [1]    ...   % replace all parent by the selected (best) offspring
                     );

% End of function