Problem with 'Logical indexing requires support of variable-sized arrays, which is currently disabled'

Hello Dears,

I have written the below code in .m file and it was OK. But for one part of my work, I should put it in Embedded function in SIMULINK, and as i run it this error appears. I would be appreciated if anybody can help me.

 for j=1:5
        if j==1 && pmuA9 >=0 && pmuA8 >=0 && pmuA7 >=0 && pmuA6 >=0 && pmuA5 >=0 && pmuA4 >=0 && pmuA3 >=0 && pmuA2 >=0 && pmuA1 >=0 
           summuA_Q1=muA_9(j)+muA_8(j)+muA_7(j)+muA_6(j)+muA_5(j)+muA_4(j)+muA_3(j)+muA_2(j)+muA_1(j);
           sumEyimuA_Q1=EyimuA_9(j)+EyimuA_8(j)+EyimuA_7(j)+EyimuA_6(j)+EyimuA_5(j)+EyimuA_4(j)+EyimuA_3(j)+EyimuA_2(j)+EyimuA_1(j);
           yPFLS_Q1=(sumEyimuA_Q1)/(summuA_Q1); 
           yPFLS_Q1(isnan(yPFLS_Q1))=0;                             
           PyQ1=[PmuA_9(j) PmuA_8(j) PmuA_7(j) PmuA_6(j) PmuA_5(j) PmuA_4(j) PmuA_3(j) PmuA_2(j) PmuA_1(j)];
           PyQ1nonZERO=nonzeros(PyQ1);
           PyQ1nonZERO(isempty(PyQ1nonZERO))=0;                     
           PyPFLS_Q1=max(PyQ1nonZERO);
        end
 end

The error is:

Logical indexing requires support of variable-sized arrays, which is currently disabled.

Function 'Embedded MATLAB Function' (#22.38943.38960), line 910, column 23:

"isnan(yPFLS_Q1)"

Answer 1

[Walter Roberson]

           PyQ1nonZERO = nonzeros(PyQ1);
           if isempty(PyQ1nonZERO)
               PyPFLS_Q1 = 0;
           else
             PyPFLS_Q1 = max(PyQ1nonZERO);
           end

[azadeh jalali]

Thanks for your response. But the error is on 'yPFLS_Q1'. The outcome of ' yPFLS_Q1=(sumEyimuA_Q1)/(summuA_Q1)' for some value is 'NAN', and I want to change 'nan' to zero. the structure of the code in .m file is correct but in Embedded function of Simulink does not answer.

[Walter Roberson]

 for j=1:5
        if j==1 && pmuA9 >=0 && pmuA8 >=0 && pmuA7 >=0 && pmuA6 >=0 && pmuA5 >=0 && pmuA4 >=0 && pmuA3 >=0 && pmuA2 >=0 && pmuA1 >=0 
           summuA_Q1=muA_9(j)+muA_8(j)+muA_7(j)+muA_6(j)+muA_5(j)+muA_4(j)+muA_3(j)+muA_2(j)+muA_1(j);
           sumEyimuA_Q1=EyimuA_9(j)+EyimuA_8(j)+EyimuA_7(j)+EyimuA_6(j)+EyimuA_5(j)+EyimuA_4(j)+EyimuA_3(j)+EyimuA_2(j)+EyimuA_1(j);
           yPFLS_Q1=(sumEyimuA_Q1)/(summuA_Q1);
           if isnan(yPFLS_Q1)
             yPFLS_Q1 = 0;
           end
           PyQ1=[PmuA_9(j) PmuA_8(j) PmuA_7(j) PmuA_6(j) PmuA_5(j) PmuA_4(j) PmuA_3(j) PmuA_2(j) PmuA_1(j)];
           PyQ1nonZERO = nonzeros(PyQ1);
           if isempty(PyQ1nonZERO)
              PyPFLS_Q1 = 0;
           else
              PyPFLS_Q1 = max(PyQ1nonZERO);
           end
        end
 end

It is not clear what the purpose of the loop is. You for j=1:5 but your if tests strictly j == 1 and there is no else for the cases where j is anything else. Therefore you might as well not use for j and instead just have j=1 and remove the j==1 from the test.

With the code only being valid for j == 1 then you should ask if it makes sense to index with j, since j is fixed: perhaps it makes more sense to put in literal 1's instead.

[Walter Roberson]

Can your PmuA_* variables ever be negative? If not then you can just max(PyQ1) instead of taking nonzeros() and max() that: if the result is 0 then there were no positive values, and if it is certain there are no negative values (hypothesis) then that would be enough to know.

[Walter Roberson]

It is confusing to see the same code have pmuA9 and PmuA_9 ? Because what I asked a moment ago about negative PmuA_* variables would be ruled out by the series of tests against pmuA* variables if they are the same variables as the PmuA_ variables.