Error "Subscript indices must either be real positive integers or logicals." At random index

I'm having some trouble with a for loop. For some reason, it works properly until the first of the 3 loops gets to the number 4.8 (when j=4.8) where it stops the program and shows the error "Subscript indices must either be real positive integers or logicals."

This is the code:

syms s;
s=tf('s');

X=zeros(25,25,25);
for i=2:0.2:7
    for a=2:0.2:7
        for j=2:0.2:7
            L=(s+i)*(s+j)*(s+a)/s/(s+10000)^2;
            Y=step(L/(1+L),t);
            if max(Y)<1.11
                X(i*5-9,5*j-9,a*5-9)=1;
            else
                X(i*5-9,5*j-9,a*5-9)=0;
            end
        end
    end
end        

(Edited)

Anyone knows what is going on?

Answer 1

[Adam Danz]

Indexing requires positive integer values just as the error message indicates. At lease one of the three indices of X is not a positive integer.

X(i*5-9,  5*j-9,  a*5-9)

[Update]

The 2nd index (5*j-9) is producing what appears to be an integer (15) but, in fact, is not an integer.

sprintf('%.20f',5*j-9)
ans =
    '15.00000000000000355271'

That's because "j" is not really 4.8

sprintf('%.20f',j)
ans =
    '4.80000000000000071054'

The reason is due to round-off error associated with floating point number representation when you create this sequence 2:0.2:7. A great article was written by a cofounder of MathWorks, Cleve Moler, that explains this problem.

Instead, use integers to control your loops.

X=zeros(25,25,25);
iSeq = 2:0.2:7; 
aSeq = 2:0.2:7; 
jSeq = 2:0.2:7; 

for i=1:numel(iSeq)
    for a=1:numel(aSeq)
        for j=1:numel(jSeq)
            L=(s+iSeq(i))*(s+jSeq(j))*(s+aSeq(a))/s/(s+10000)^2;
            Y=step(L/(1+L),t);
            if max(Y)<1.11
                X(i,j,a)=1; % <-- double check that this is correct
            else
                X(i,j,a)=0; % <-- double check that this is correct
            end
        end
    end
end 

Also, this will surely result in an error

 L=*(s+i)*(s+j)*(s+a)/s/(s+10000)^2;
 % ^

[Andibadia]

Thank you! The last part was a mistake copying the code from MATLAB. But yes, I used the function round() before each number and now it works great!

[Adam Danz]

OK! I updated my answer to suggest an alternative but if round() works, then that would be OK, too.