How can I this code working

function [al, aa, au fl, fa, fu, nFuncCall]=GoldenSectionPhaseI(func,X0,d,delta)

q=0;

a1=0;

a2=delta;

nFuncCall=0;

F1=func(X0+a1*d);

nFuncCall=nFuncCall+1;

F2=func(X0+a2*d);

nFuncCall=nFuncCall+1;

while(1)

q=q+1;

a3=0;

for j=0:q

a3=a3+(1.618)^j*delta;

end

F3=func(X0+a3*d);

if (F1>F2 && F2<F3)

break

else

a1=a2; F1=F2;

a2=a3; F2=F3;

end

end

al=a1;

aa=a2;

au=a3;

fl=F1;

fa=F2;

fu=F3;

end

[Dyuman Joshi]

What seems to be the problem?

And how are you calling the function? What are the input values?

[Sam Chak]

Looks like the Golden Ratio section root-finding technique. Perhaps OP wants to test it out on some nonlinear functions, but doesn't know how to call the function.

Answer 1

[Divit]

Hi Seojin,

I understand that you would like to test and run the code you provided.

The code you provided implements the Golden Section Search method for finding the minimum of a unimodal function. However, there are a few syntax errors and missing variable declarations in the code you provided. Here's the corrected code:

Here's how you can use this function:

1. Define your objective function "func" that takes a vector input and returns a scalar output.
2. Define your initial point "X0" as a vector.
3. Define the search direction "d" as a vector.
4. Define the initial step size "delta".
5. Call the "GoldenSectionPhaseI" function with the appropriate arguments.

You can refer to the following example code:

% Define the objective function
func = @(x) 100 * (x(2) - x(1)^2)^2 + (1 - x(1))^2;  % Example objective function: Rosenbrock's function(replace with your own)

% Define the initial point and search direction as 2-dimensional vectors
X0 = [0; 0];   % Initial point
d = [1; 1];    % Search direction

% Define the initial step size
delta = 0.1;

% Call the GoldenSectionPhaseI function
[al, aa, au, fl, fa, fu, nFuncCall] = GoldenSectionPhaseI(func, X0, d, delta);

% Display the results
disp(['al: ', num2str(al)]);

al: 0.52359

disp(['aa: ', num2str(aa)]);

aa: 0.94717

disp(['au: ', num2str(au)]);

au: 1.6325

disp(['fl: ', num2str(fl)]);

fl: 6.4492

disp(['fa: ', num2str(fa)]);

fa: 0.25316

disp(['fu: ', num2str(fu)]);

fu: 107.0291

disp(['nFuncCall: ', num2str(nFuncCall)]);

nFuncCall: 2

function [al, aa, au, fl, fa, fu, nFuncCall] = GoldenSectionPhaseI(func, X0, d, delta)
q = 0;
a1 = 0;
a2 = delta;
nFuncCall = 0;
F1 = func(X0 + a1 * d);
nFuncCall = nFuncCall + 1;
F2 = func(X0 + a2 * d);
nFuncCall = nFuncCall + 1;

while true
    q = q + 1;
    a3 = 0;
    for j = 0:q
        a3 = a3 + (1.618)^j * delta;
    end
    F3 = func(X0 + a3 * d);
    
    if (F1 > F2 && F2 < F3)
        break
    else
        a1 = a2;
        F1 = F2;
        a2 = a3;
        F2 = F3;
    end
end

al = a1;
aa = a2;
au = a3;
fl = F1;
fa = F2;
fu = F3;
end

Feel free to replace the objective function with your own function that you want to minimize using the Golden Section Search method. Make sure to adjust the dimensions of "X0" and "d" based on your specific optimization problem. For example, if your objective function takes a vector of two variables, you would need to set "X0" and "d" as 2-dimensional vectors.

I hope it helps!