[Val] = ComputeSchottsSpacingMetric(Pop) ComputeSchottsSpacingMetric: Computes the Schotts Spacing Metric for a Pareto set approximation INPUTS Pop: Each row corresponds to a vector OUTPUTS Val: Average square difference between pairs of vectors Last version 2/26/2009. Roberto Santana (roberto.santana@ehu.es)
0001 function[Val] = ComputeSchottsSpacingMetric(Pop) 0002 % [Val] = ComputeSchottsSpacingMetric(Pop) 0003 % ComputeSchottsSpacingMetric: Computes the Schotts Spacing Metric for a 0004 % Pareto set approximation 0005 % INPUTS 0006 % Pop: Each row corresponds to a vector 0007 % OUTPUTS 0008 % Val: Average square difference between pairs of vectors 0009 % 0010 % Last version 2/26/2009. Roberto Santana (roberto.santana@ehu.es) 0011 0012 Val = 0; 0013 [PopSize,n] = size(Pop); 0014 di = n*max(max(Pop))*ones(1,PopSize); 0015 npairs = PopSize*(PopSize-1)/2; 0016 for i=1:PopSize, 0017 for j=1:PopSize, 0018 if i~=j 0019 di(i) = min(di(i),sum(abs(Pop(i,:)-Pop(j,:)))); 0020 end 0021 end 0022 end 0023 d = mean(di); 0024 0025 Val = sqrt((1/(PopSize-1))*sum( (di-d).^2)); 0026 0027