Using subs() in a function when symbolic variables defined in another function

Hello Everyone! I'm having some problems using the subs() function and I need your help.

I defined a function to calculate a jacobian matrix using symbolic variables. For example:

function myJacobian = getJacobian()

    % Define function variables
    syms x y z  
    
    % Define myFunction
    myFunction = x^2 + y^2 + z^2;
    
    % Calculate myJacobian
    myJacobian = jacobian(myFunction, [x,y,z]);
    
end

Now, in "main", I am calling this function and obtaining the jacobian. Then, I am substituting the symbolic variables (x,y,z) with some numbers in another function

% Get the jacobian matrix F
F = getJacobian;

% Get my result
value_F = someRandomFunc(F, someOtherInputs);

where "someRandomFunc" is given as

function value_F = someRandomFunc(F)

    currentX = getX(someRandomInputs1);
    currentY = getY(someRandomInputs2);
    currentZ = getZ(someRandomInputs3);
    
    value_F = subs(F, {x,y,z}, {currentX,currentY,currentZ});
end

However, I am getting the following error

"Unrecognized function or variable 'x'."

I know that I am getting this error because these symbolic variables are not defined in the workspace of "someRandomFunc". So my question is that it is possible to transfer these symbolic variables defined in "getJacobian()" function to "someRandomFunc" so that I can use subs() with complacency. I tried to simplify my functions so that you can understand the problem better. My "somRandomFunc" function is in a for loop. At every iteration I am obtaining different x,y,z values and calculating the jacobian for this specific iteration accordingly. I do not want to define symbolic variables in "someRandomFunc" since it slows down the program a lot. Thanks for the help in advance.

Kind regards.

Answer 1

[Paul]

Making F into a symfun seems to work ....

F = getJacobian

F(x, y, z) = 

valueF = someRandomFunc(F)

valueF = 

function myJacobian = getJacobian()
    % Define function variables
    syms x y z  
    
    % Define myFunction
    myFunction = x^2 + y^2 + z^2;
    
    % Calculate myJacobian
    myJacobian(x,y,z) = jacobian(myFunction, [x,y,z]);
    
end

function value_F = someRandomFunc(F)
    currentX = 1;
    currentY = 2;
    currentZ = 3;
    
    value_F = F(currentX,currentY,currentZ);
end

[hknatas]

Hello Paul!

Thank you for your quick answer. Error has been disappeared, however, now I have an another problem. The class of "value_F" becomes 'sym'. Is there a way to simply convert "value_F" to a normal matrix? Since I am using "value_F" for my other computations, it also makes the class of my subsequent results 'sym'. Please see the following images.

[hknatas]

OK, I solve the problem by simply

value_F = double(value_F)

Thank you for your answer again, I did not know that there is something like 'symfun'.

[Paul]

I thought you wanted to stay in the sym realm. If you only want a numerical calculation of the Jacobian, then consider returning it as an anonymous function. Note that F is vectorized.

F = getJacobian

F = function_handle with value:
    @(x,y,z)[x.*2.0,y.*2.0,z.*2.0]

valueF = someRandomFunc(F)

valueF = 1×3

2 4 6

function myJacobian = getJacobian()
    % Define function variables
    syms x y z  
    
    % Define myFunction
    myFunction = x^2 + y^2 + z^2;
    
    % Calculate myJacobian
    myJacobian = matlabFunction(jacobian(myFunction, [x,y,z]),'Vars',{'x' 'y' 'z'});
    
end
function value_F = someRandomFunc(F)
    currentX = 1;
    currentY = 2;
    currentZ = 3;
    
    value_F = F(currentX,currentY,currentZ);
end

[hknatas]

Yes, what I was looking for was matlabFunction(). Thanks a lot!