"User supplied objective function must return a scalar value." error using fminbnd

Hello everyone. I am trying to implement the steepest descent algorithm to a convex and nonquadratic function "fun". I need to find the a (alpha) by using fminbnd function in the interval [1,10]. However, I am getting the error "Error using fminbnd: User supplied objective function must return a scalar value." Since xnew and dk matrices also change everytime while loop runs, I could not define the xnew2 function as function handle (I guess, I am not quite sure about the definition of function handle.) I would be glad if someone could help me out with this problem. Thanks a lot.

syms fun(x,y)  xnew2(a)
fun(x,y) =(18.*y-0.2.*(x-20).^2).^2 + (20.*x-0.1.*(y-10).^2).^2; %function is defined
g=(gradient(fun,[x,y])); %gradient vector
x0=[0.2 0.2]'; %initial points
xnew=x0; %I don't want to change initial points but I need another variable to start with
iter=0;
dk=-g(xnew(1),xnew(2));

while(norm(dk)>1e-05)
    iter=iter+1; 
    xnew2(a)= xnew+a*dk; %xnew2 is defined with respect to a which as unknown for now

    a_new=fminbnd(xnew2,1,10); %a_new will be the local min in the interval 1,10 
    xnew=xnew+a_new*dk; %xnew will get a new value with the variable
   
    X=['The x vector at iteration ',num2str(iter)];
    disp(X)
    display(double(xnew));
    
    dk=-g(xnew(1),xnew(2));
end

Error using fminbnd
User supplied objective function must return a scalar value.

Answer 1

[Torsten]

syms a
x = sym('x',[2,1]);
fun(x) =(18.*x(2)-0.2.*(x(1)-20).^2).^2 + (20.*x(1)-0.1.*(x(2)-10).^2).^2; %function is defined
g = gradient(fun); %gradient vector
x0 = [0.2 0.2]'; %initial points
xnew = x0; %I don't want to change initial points but I need another variable to start with
iter = 0;
dk = -g(xnew(1),xnew(2));
while norm(dk) > 1e-05 && iter < 9
    iter = iter+1; 
    xnew2 = xnew+a*dk; %xnew2 is defined with respect to a which as unknown for now
    f = matlabFunction(fun(xnew2(1),xnew2(2)));

    a_new = fminbnd(f,0,10) %a_new will be the local min in the interval 1,10 
    xnew = xnew+a_new*dk; %xnew will get a new value with the variable
   
    %X=['The x vector at iteration ',num2str(iter)];
    %disp(X)
    %display(double(xnew));
    
    dk = -g(xnew(1),xnew(2));
    double(xnew)
    double(fun(xnew(1),xnew(2)))
end

a_new = 0.0010

ans = 2×1

1.6416 2.9776

ans = 969.1248

a_new = 0.0016

ans = 2×1

0.1870 3.6446

ans = 166.7014

a_new = 0.0010

ans = 2×1

0.4113 4.1267

ans = 28.8816

a_new = 0.0017

ans = 2×1

0.1559 4.2564

ans = 4.6244

a_new = 0.0010

ans = 2×1

0.1976 4.3348

ans = 0.7119

a_new = 0.0017

ans = 2×1

0.1578 4.3565

ans = 0.1070

a_new = 0.0010

ans = 2×1

0.1642 4.3684

ans = 0.0163

a_new = 0.0017

ans = 2×1

0.1581 4.3718

ans = 0.0024

a_new = 9.9955e-04

ans = 2×1

0.1591 4.3735

ans = 3.4421e-04

[Nilgün Uzun]

Thank you soo much it worked perfectly :)