How to use 'solve' function in simulink , using matlab function block

I am trying to use "solve' function inside simulink making use of the Matlab function block. The function is as given below

function y = fcn(u)
%#codegen
coder.extrinsic('syms');

P=u;
V=2*(P*0.0031-pi);

syms a value  ;
value = V;

sol=solve((sin(2*a)-(2*a))== value);

y=0;
y = sol;

it gives an error.. " undefined function or variable 'a' "

Answer 1

[Walter Roberson]

You could change

   syms a value  ;

to

a = sym('a');

and you could change

sol=solve((sin(2*a)-(2*a))== value);

to

sol = double( solve( ((sin(2*a)-(2*a))) - (value), a) );

However, you have a more fundamental problem. Embedded MATLAB Function blocks are for code that is to be optimized by compiling, and it is not allowed to compile anything in the Symbolic Toolbox.

You are going to need to switch to fzero() or the like. The solution is positive for P < pi/*0.0031 and negative for larger P, which might help you decide on the appropriate bounds to use.

[Nithin S]

Thanks for the reply, as per your suggestion I am trying to use fzero() now, I have modified the function with fzero(), the value of 'u' will be in the range of [100 1000], for this I manually calculated the value of 'a' as x0=[2.973 0.40089 ] and set is as interval.

function y=fcn(u)

V=2*(u*0.0031-pi)  ;

myfun = @(a)(sin(2*a)-(2*a) - V);  

x0=[2.3322 0.40089 ] ;            
y=0;    
y = fzero(myfun,x0)

during execution I am getting an error

This kind of expression is not supported.

Function 'MATLAB Function' (#33.381.405), line 19, column 9:
"@(a)(sin(2*a)-(2*a) - V)"

when I run the script in workspace it is working. what is that I am doing wrong ?

[Walter Roberson]

Anonymous functions are not supported, I think. You would need to create a different true function that did the calculation for you. According to the documentation you would use a handle to that function. I think you should be able to put it in the same file.

I am not sure if nested functions are supported. To share parameters you might possibly need to resort to global. Perhaps

function y = fcn(u)

V = 2*(u*0.0031-pi);

global myfun_V
myfun_V = V;

y = 0;
a0=[0.40089 2.3322] ; 
y = fzero(@myfun, a0);

function residue = myfun(a)
global myfun_V
residue = 0;
residue = (sin(2*a)-(2*a) - myfun_V);