Documentation of recmp

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

Function Synopsis

NewChrom = recmp(OldChrom, RecRate, RecXpt, RecRs);

Help text

 RECombination Multi-Point, low level function

 This function takes a matrix OldChrom containing the 
 individuals in the current population, applies recombination
 to consecutive pairs of individuals with probability RecRate 
 and returns the resulting offspring.
 The number of recombination points to use and the forced 
 production of offspring different from their parents can be
 parametrized.
 Every representation of the variables can be used.
 This function is a low level function used by the corresponding
 higher level function, for instance: recsp, recdp, recshrs.

 Syntax:  NewChrom = recmp(OldChrom, RecRate, RecXpt, RecRs)

 Input parameters:
    OldChrom  - Matrix containing the chromosomes of the old
                population. Each row corresponds to one individual
                (in binary form).
    RecRate   - Probability of recombination ocurring between pairs
                of individuals.
    RecXpt    - Scalar indicating the number of crossover points
                1: single point recombination
                2: double point recombination
                0: shuffle point recombination
    RecRs     - reduced surrogate
                0: no reduced surrogate
                1: reduced surrogate

 Output parameter:
    NewChrom  - Matrix containing the offspring after mating, 
                ready to be mutated and/or evaluated, in the same 
                format as OldChrom.

 See also: recombin, recdp, recsh, recsp, recdprs, recsprs, recshrs

Cross-Reference Information

Listing of function recmp

% Author:   Hartmut Pohlheim
% History:  31.05.95    file created
%           24.07.97    calculation of Mask changed for compatibility
%                       with Matlab5, see MaskIsZero
%           07.04.98    names of parameters changed


function NewChrom = recmp(OldChrom, RecRate, RecXpt, RecRs);

% Identify the population size (Nind) and the chromosome length (Nvar)
   [Nind, Nvar] = size(OldChrom);

% Set default parameters
   if Nvar < 2, NewChrom = OldChrom; return; end

   if nargin < 4, RecRs = []; end
   if nargin < 3, RecXpt = []; end
   if nargin < 2, RecRate = []; end
   if isnan(RecRate), RecRate = []; end
   if isnan(RecXpt), RecXpt = []; end
   if isnan(RecRs), RecRs = []; end
   if isempty(RecRate), RecRate = 0.7; end
   if isempty(RecXpt), RecXpt = 0; end
   if isempty(RecRs), RecRs = 0; end

   Xops = floor(Nind/2);
   DoCross = rand(Xops,1) < RecRate;
   odd = 1:2:Nind-1;
   even = 2:2:Nind;

% Compute the effective length of each chromosome pair
   Mask = ~RecRs | (OldChrom(odd, :) ~= OldChrom(even, :));
   Mask = cumsum(Mask')';
   % Check for Mask with all zero ==> both individuals are identical
   % set values to one to remove warning divide by zero in xsites(:,2)= ...
   MaskIsZero = find(Mask(:, Nvar)==0);
   Mask(MaskIsZero,:) = ones(length(MaskIsZero), size(Mask, 2));
   % if any(Mask(:, Nvar)==0), Mask(:, Nvar), end

% Compute cross sites for each pair of individuals, according to their
% effective length and RecRate (two equal cross sites mean no recombination)
   xsites(:, 1) = Mask(:, Nvar);
   if RecXpt >= 2,
      xsites(:, 1) = ceil(xsites(:, 1) .* rand(Xops, 1));
   end
   xsites(:,2) = rem(xsites + ceil((Mask(:, Nvar)-1) .* rand(Xops, 1) ) ...
                     .* DoCross - 1 , Mask(:, Nvar) ) + 1;

% Express cross sites in terms of a 0-1 mask
   Mask = (xsites(:,ones(1,Nvar)) < Mask) == (xsites(:,2*ones(1,Nvar)) < Mask);

   if ~RecXpt,
      shuff = rand(Nvar,Xops);
      [ans,shuff] = sort(shuff);
      for i=1:Xops
         OldChrom(odd(i),:)=OldChrom(odd(i),shuff(:,i));
         OldChrom(even(i),:)=OldChrom(even(i),shuff(:,i));
      end
   end

% Perform recombination
   NewChrom(odd,:)  = (OldChrom(odd,:) .*   Mask)  + (OldChrom(even,:) .* (~Mask));
   NewChrom(even,:) = (OldChrom(odd,:) .* (~Mask)) + (OldChrom(even,:) .*   Mask);

% If the number of individuals is odd, the last individual cannot be mated
% but must be included in the new population
   if rem(Nind,2), NewChrom(Nind,:)=OldChrom(Nind,:); end

   if ~RecXpt,
      [ans,unshuff] = sort(shuff);
      for i=1:Xops
         NewChrom(odd(i),:)=NewChrom(odd(i),unshuff(:,i));
         NewChrom(even(i),:)=NewChrom(even(i),unshuff(:,i));
      end
   end


% End of function