How to use fminsearch for a function converted from symbolic expression

Hello,

I am trying to fit a plane to a set of data. I'd like to write my own function with fminsearch function.

I created the objective function using symbolic, and then I convert the symbolic expression to a function handle using "matlabFunction". Finally, I pass the function handle to the fminsearch function. However, Matlab reports "Not enough input arguments" How do I fix this?

My code is as the following

% I have 36 3D points. I'd like to use a plane to fit them
% Pcr_th is a 36-3 matrix that contains the point coordinates, the columns represent X, Y and Z coordinates

syms A B C D
sum_sqd = 0;
for i = 1:36
    d =  ( A*Pcr_th(i,1) + B*Pcr_th(i,2) +C*Pcr_th(i,3) + D) / (sqrt(A*A+B*B+C*C));
    sqd = d*d;
    sum_sqd = sum_sqd + sqd;
end

% find the centroid
ctroid_x = sum(Pcr_th(i,1))/36;
ctroid_y = sum(Pcr_th(i,2))/36;
ctroid_z = sum(Pcr_th(i,3))/36;

% initial guess for D
Dint = -(ctroid_x+ ctroid_y + ctroid_z);

% Create the objective function
fun = matlabFunction(sum_sqd,'vars',[A,B,C,D]);
intg = [1, 1, 1, Dint];  % this is the initial guess.
a = fminsearch (fun, intg)  % This reports Not enough input arguments. And error in fun=matlabFunction...

Answer 1

[Walter Roberson]

fun = matlabFunction(sum_sqd, 'vars', {[A,B,C,D]});

When you pass in a vector of variables for the 'vars' option, then matlabFunction uses one input parameter for each variable.

When you pass in a cell array for the 'vars' option, then matlabFunction uses one input parameter for each element of the cell array, packing all of the variables present in that one entry into a single parameter.