Hello, I have a very basic doubt regarding genetic algorithm. GA gives different answers each time we run it and so how can we rely on it. I mean to ask how far can we take the results for granted. I googled about it and came across using 'optimoptions' and 'gs' in order to see that the solution approaches global minima rather than local, but could not understand even a bit. Can anyone help me in understanding that? How do we code it in Matlab?
Following is the code I have written in Matlab to minimise a function defined in square brackets (called as 'calculations' in the code). This code, as told above returns different values every time I execute it. How do I refine it for precise results?
%defining our search space
x=linspace(0,0.6,150);
y=linspace(0,0.6,150);
%creating 2d mesh
[xx yy]=meshgrid(x,y);
%evaluation
for i=1:length(xx)
for j=1:length(yy)
inputvector(1)=xx(i,j);
inputvector(2)=yy(i,j);
f(i,j)=calculations(inputvector);
end
end
%plotting the inputs vs function
surfc(xx,yy,f)
xlabel('xvalues');
ylabel('yvalues');
zlabel('functionvalues');
%getting a shaded plot rather the one with grids
shading interp
%optimising the funtion
[inputs_of_optimal_solution,optimal_fvalue]=ga(@calculations,2)
disp('The f minima is')
disp(optimal_fvalue)
disp('The input values corresponding to f minima are')
disp(inputs_of_optimal_solution)
[
function [actualfunc]=calculations(inputvector)
f1x=(sin((5.1*pi*inputvector(1))+0.5))^6;
f1y=(sin((5.1*pi*inputvector(2))+0.5))^6;
f2x=exp((-4*log(2)*(inputvector(1)-0.0667)^2)/0.64);
f2y=exp((-4*log(2)*(inputvector(2)-0.0667)^2)/0.64);
actualfunc=f1x*f2x*f1y*f2y;
end
]
Hi Kiran,
First I propose you to see answer of similar question posed through ResearchGate here is the link :
- What results should I pic up if my genetic algorithm is... - ResearchGate
https://www.researchgate.net/.../What_results_should_I_pic_up_if_...
Also I suggest you to see links and attached files in topic.
-Why Genetic Algorithm gives different results for optimization of ...
https://stackoverflow.com/.../why-genetic-algorithm-gives-differen...
-How the Genetic Algorithm Works - MATLAB & Simulink - MathWorks
https://www.mathworks.com › ... › Genetic Algorithm
-How does one verify the results of simulation from genetic algorithm ...
https://www.mathworks.com/matlabcentral/answers/146344-how-d...
-Linear Algebra and Probability for Computer Science Applications
https://books.google.dz/books?isbn=1466501553
Best regards
Reset MATLAB before each run, you will get exactly the same result. One way to reset MATLAB is to close and open again.
I came to know that the recent versions of Matlab facilitate with commands (optimoptions) which take our results closer to the global minima. But, the older ones like 2013a (which I use) do not have such options. May be we can define the search space and limit it using upper and lower boundaries so that GA gives us atleast a reliable result, if not the perfect solution. But still how can we rely on on the results and consider them for our applications?
If same seeds of random numbers will be used in each run then the results will be same by any heuristic algorithm including GA. It also no matter which programming environment is used.
“RNG” command of Matlab which gives us the same realization every execution). In fact, RNG('default') puts the settings of the random number generator used by RAND, RANDI, and RANDN to their default values so that they produce the same random numbers as if you restarted MATLAB.
Yes, this problem is very common in Genetic algorithm, this is due to entrap of local search and this occurs due to premature convergence of the chromosomes. I suggest you to use fuzzy roulette wheel selection criteria in which the premature convergence of the chromosomes is being handled.
How can i solve this and whocan help me i am using Matlab GA toolbox and i have got teh following errors:
Obj =
1.0e+13 *
2.3579 + 0.0097i 0.0106 + 0.0000i
Obj =
1.0e+11 *
4.0416 + 2.4312i 0.0186 + 0.0110i
Obj =
1.0e+12 *
1.8716 + 0.2003i 0.0085 + 0.0009i
Warning: One or more feasible individuals has a complex fitness function value. gamultiobj is using the real part of the
fitness function values
> In rankAndDistance (line 21)
In gamultiobjMakeState (line 199)
In gamultiobjsolve (line 20)
In gamultiobj (line 304)
In Main_Genetic_Algorithm (line 41)
Warning: Imaginary parts of complex X and/or Y arguments ignored.
> In gaplotpareto>plotFirstFront (line 72)
In gaplotpareto (line 28)
In gadsplot>callOnePlotFcn (line 194)
In gadsplot (line 141)
In gamultiobjsolve (line 24)
In gamultiobj (line 304)
In Main_Genetic_Algorithm (line 41)
Warning: Using only the real component of complex data.
> In matlab.graphics.chart.internal.getRealData (line 52)
In bar (line 139)
In gaplotscores (line 41)
In gadsplot>callOnePlotFcn (line 194)
In gadsplot (line 141)
In gamultiobjsolve (line 24)
In gamultiobj (line 304)
In Main_Genetic_Algorithm (line 41)
Error using matlab.graphics.chart.primitive.Line/set
Complex values are not supported.
Error in gaplotpareto>plotFirstFront (line 84)
set(plotHandle,'Xdata',xy(:,1), 'Ydata',xy(:,2));
Error in gaplotpareto (line 28)
plotFirstFront(state.Score(range,:),state.Rank(range),markers{1 + mod(i,5)},objectivesToPlot(:)',flag,tag);
Error in gadsplot>callOnePlotFcn (line 194)
optimvalues = plotfcn(varargin{1:end});
Error in gadsplot (line 148)
[state,optimvalues] =
callOnePlotFcn(fname,plotNames{i},state,options.OutputPlotFcnOptions,optimvalues,flag,args{i}{:});
Error in gamultiobjsolve (line 64)
state = gadsplot(options,state,currentState,'Genetic Algorithm');
Error in gamultiobj (line 304)
[x,fval,exitFlag,output,population,scores] = gamultiobjsolve(FitnessFcn,nvars, ...
Error in Main_Genetic_Algorithm (line 41)
[X, fval, exitflag, output, population, scores] = gamultiobj(@Fitness_Function, ...
Obj =
1.0e+11 *
5.5973 + 1.6902i 0.0256 + 0.0076i
Obj =
1.0e+11 *
2.7293 + 3.0808i 0.0127 + 0.0139i
Obj =
1.0e+11 *
5.0123 + 1.5909i 0.0229 + 0.0072i
Obj =
1.0e+11 *
4.0736 + 1.6777i 0.0187 + 0.0076i
Obj =
1.0e+11 *
8.2966 + 1.5221i 0.0377 + 0.0069i
Obj =
1.0e+11 *
6.3667 + 1.6824i 0.0290 + 0.0076i
- Badges
- Science method