Specify initial points (both variable and objective function) in surrogateopt function for some data

Hello,

I would like to add some initial data (which are from previous run of the objective function) to surrogateopt to reduce the computation time. The objective function have 4 variable. I have 100 previously run data of the objective function.

Below is a simple example of the code:

objconstr = @(x)(x(:,1)^2+x(:,2)^2+x(:,3)^2+x(:,4)^2);

LB = [-2,-6, -3, -5];

UB = -LB;

options = optimoptions('surrogateopt','PlotFcn');

x0=[0 0 0 0;

1 1 1 1;

......

];

%Results of objective function for the x0 values

y0=[1 3 4 5 ....]

[x,fval,exitflag,output] = surrogateopt(objconstr,LB,UB,options)

[Torsten]

I would like to add some initial data (which are from previous run of the objective function) to surrogateopt to reduce the computation time.

Did you read somewhere that such an option exists ? I cannot find it.

[Mohammad Azadi Tabar]

Hello,

Yes, I did and found the answer.

trials.X=x0;

trials.Fval=y0;

% options.InitialPoints = trials;

options = optimoptions('surrogateopt','PlotFcn','surrogateoptplot',...

'MinSampleDistance',1e-02,...

'MaxFunctionEvaluations',10,'InitialPoints', trials);

[x,fval,exitflag,output] = surrogateopt(objconstr,LB,UB,options)