Conversion to struct from double is not possible.

Hello

I have problem with my program

Here' the code

function f=funkcja1(x)
f= 8010+5*x(4)^2+40*x(2)*x(4)+40*x(1)*x(4)+6*x(3)^2+84*x(2)*x(3)+379*x(2)^2+310*x(1)*x(2)+1224*x(1)^2+2712*x(1)+10*x(4)-84*x(3)-418*x(2);

function gauss(xp,epsilon)
d=eye(4);
syms x1 x2 x3 x4 alfa;


fout=fopen('metgauss_przebieg_nowy.txt','w');
fprintf(fout,'Punkt poczatkowy: %10s %10s %10s %10s\r\n', num2str(xp(1,1)),num2str(xp(1,2)),num2str(xp(1,3)),num2str(xp(1,4)));
fprintf(fout,'%10s %43s %10s %10s\r\n','Iteracja','Punkt','f(x)','kryterium stopu');

fprintf(fout,'%10d %10s %10s %10s %10s %10s %10s\r\n',0,num2str(xp(1,1)),num2str(xp(1,2)),num2str(xp(1,3)),num2str(xp(1,4)),num2str(funkcja(xp)), '0');
tic
x0=xp;

for i=1:4
	[a, b, c, z]= fminsearch(@funkcja1,[1 2 3 4]);

     xp=xp+[a, b, c, z]*d(i,:);
end

iteracja=1;
while norm(xp-x0) >= epsilon
    fprintf(fout,'%10d %10s %10s %10s %10s %10s %10s\r\n',iteracja,num2str(xp(1,1)),num2str(xp(1,2)),num2str(xp(1,3)),num2str(xp(1,4)),num2str(funkcja(xp)), num2str(norm(xp-x0)));
    fprintf('Interacja=%d   Kryterium stopu=%d\n',iteracja, norm(xp-x0) );
	x0=xp;
    for i=1:4
		[a,b,c,z]  = fminsearch(@funkcja1,[1 2 3 4]);
        xp=xp+[a,b,c,z]*d(i,:);
    end
    iteracja=iteracja+1;
    if iteracja==100
 disp('Iteracja przekroczyla milion');
 break
    end
end

[Konrad]

Hi,

did you check the man page of fminsearch()?

[x,fval,exitflag,output] = fminsearch(___)

The 4th ouput 'output' is a structure. Although I don't know what your code is supposed to do, the way you use the outputs of fminsearch (a,b,c,z) doesn't make much sense to me...

Also, you did not mention how you call your function, i.e. what is xp and epsilon?

Btw, formatting would make your question easier to read.

[Michal Kopysc]

Sorry for my post but i am not good at that website so i copy paste my code, here is my all code, this program should search min of function

function f=funkcja1(x)
f= 8010+5*x(4)^2+40*x(2)*x(4)+40*x(1)*x(4)+6*x(3)^2+84*x(2)*x(3)+379*x(2)^2+310*x(1)*x(2)+1224*x(1)^2+2712*x(1)+10*x(4)-84*x(3)-418*x(2);

function gauss(xp,epsilon)
d=eye(4);
syms x1 x2 x3 x4 alfa;


fout=fopen('metgauss_przebieg_nowy.txt','w');
fprintf(fout,'Punkt poczatkowy: %10s %10s %10s %10s\r\n', num2str(xp(1,1)),num2str(xp(1,2)),num2str(xp(1,3)),num2str(xp(1,4)));
fprintf(fout,'%10s %43s %10s %10s\r\n','Iteracja','Punkt','f(x)','kryterium stopu');

fprintf(fout,'%10d %10s %10s %10s %10s %10s %10s\r\n',0,num2str(xp(1,1)),num2str(xp(1,2)),num2str(xp(1,3)),num2str(xp(1,4)),num2str(funkcja(xp)), '0');
tic
x0=xp;

for i=1:4
	[a, b, c, z]= fminsearch(@funkcja1,[1 2 3 4]);

     xp=xp+[a, b, c, z]*d(i,:);
end

iteracja=1;
while norm(xp-x0) >= epsilon
    fprintf(fout,'%10d %10s %10s %10s %10s %10s %10s\r\n',iteracja,num2str(xp(1,1)),num2str(xp(1,2)),num2str(xp(1,3)),num2str(xp(1,4)),num2str(funkcja(xp)), num2str(norm(xp-x0)));
    fprintf('Interacja=%d   Kryterium stopu=%d\n',iteracja, norm(xp-x0) );
	x0=xp;
    for i=1:4
		[a,b,c,z]  = fminsearch(@funkcja1,[1 2 3 4]);
        xp=xp+[a,b,c,z]*d(i,:);
    end
    iteracja=iteracja+1;
    if iteracja==100
 disp('Iteracja przekroczyla milion');
 break
    end
end

[Michal Kopysc]

i dont know how can i exaclty use fminsearch

Answer 1

[Yongjian Feng]

Check

help fminsearch

The last output (called z in your script) is a struct. So line 16 doesn't make sense.

Answer 2

[Sulaymon Eshkabilov]

The error is coming from your problem exercise fcn @funkcja1. That would be helpful to see funkcja1 to give you proper hints how to fix a prompted error. Note that fminsearch() gives you four outputs, which are: [x,fval,exitflag,output]

x is found solutions

fval is the computed fcn values at x

exitflag is showing whether search suceeded or not (0 or 1)

output is a struct type of variable that contains: iterations, funcCount, algorithm, message. Where iterations are the nuber of interation, and funcCount is also number. And algorithm abd message are character strings. Thus, the way you are treating these are not correct. If you want to use the found solutions, then you should consider to use your variable "a" only.

See this documentation that explains all of these points in details: https://www.mathworks.com/help/matlab/ref/fminsearch.html

[Michal Kopysc]

function gauss(xp,epsilon)
d=eye(4);
syms x1 x2 x3 x4 alfa;


fout=fopen('metgauss_przebieg_nowy.txt','w');
fprintf(fout,'Punkt poczatkowy: %10s %10s %10s %10s\r\n', num2str(xp(1,1)),num2str(xp(1,2)),num2str(xp(1,3)),num2str(xp(1,4)));
fprintf(fout,'%10s %43s %10s %10s\r\n','Iteracja','Punkt','f(x)','kryterium stopu');

fprintf(fout,'%10d %10s %10s %10s %10s %10s %10s\r\n',0,num2str(xp(1,1)),num2str(xp(1,2)),num2str(xp(1,3)),num2str(xp(1,4)),num2str(funkcja(xp)), '0');
tic
x1=[1 2 3 4];
x0=xp;

for i=1:4
	 xmin= fminsearch(@funkcja1,x1);

     xp=xp+xmin*d(i,:);
end
________________________________________________________________________________________
Error using  * 
Incorrect dimensions for matrix multiplication. Check that the number of columns in the first matrix matches the
number of rows in the second matrix. To perform elementwise multiplication, use '.*'.

Error in gauss (line 18)
     xp=xp+xmin*d(i,:);

[Michal Kopysc]

function f=funkcja(x)
f= 1484*x(1) - 252*x(2) - 168*x(1)*x(2) - 28*x(1)^2*x(2) + 757*x(1)^2 + 168*x(1)^3 + 14*x(2)^2 + 14*x(1)^4 + 1333;

[Michal Kopysc]

here is that function

[Michal Kopysc]

the size of the table does not match

Answer 3

[Sulaymon Eshkabilov]

Use elementwise operation:

...
 xp=xp+xmin.*d(i,:);   % Elementwise
 
 end