Indexing error in function code.

I have the following function:

 function [ r ] = ost_approx_rate_conv( h_1,B )

 n = 1:length(B);
 k = 2:1:length(B);
 h = zeros(size(n));

 h(k+1) = h(k-1)/2;

 h(1) = h_1;
 h = logical(h);

 r(h) = log((B(2*h)-B(2*h))/(B(h)-B(2*h)))/log(2);

and when I try to run it I get "Subscript indices must either be real positive integers or logicals." error for my r(h) term. How exactly would I remedy this?

[dpb]

Need to first define what it is you're trying to do. This is peculiar at best and undoubtedly not what you intend but I can't tell what is wanted from the code alone.

For example, k is vector of integers from 2:length(B) and h is a zero-valued vector of length(B) but then you assign the 3rd thru (N+1)th positions to the values from the 1:(N-1)th (dividing by 2, but zero/2 is still zero, so all you've really done in the end is

 h=zeroes(1,length(B)+1);
 h(1)=h_1;

after all that.

The explicit cause of the error, however, is that when you write

B(2*h)

despite have previously converted h to logical array, that you've subsequently multiplied by 2 means its numeric and most (if not all) the entries are zero. Matlab arrays are one-based so 0 is not either a positive integer nor a logical.

BTW, the expression

log((B(2*h)-B(2*h))

is log(0) as the two terms are identical so this also is clearly not what you're intending but I've no kluw what that might be...

Answer 1

[Andrew Newell]

You'll get the same error with this code snippet:

h = logical([1 0]);
B(2*h)

or, for that matter, just

B(0)

The problem is that by multiplying h by 2, you're converting it back to a double precision number. So you have indices

2*h

ans =

     2     0

and the second index is zero. As to how you might fix it, I can't suggest anything because I don't understand what you are trying to do.

Answer 2

[Michael Haderlein]

You initialize h as zeros. k goes from 2 to length(B). The line

h(k+1) = h(k-1)/2;

remains all h as zeros (and I'm questioning if this line is doing what you want - extending h by one zero).

The line

 h = logical(h);

sets all these zeros (only the first element of h might have been set nonzero) to false. However, the line

 r(h) = log((B(2*h)-B(2*h))/(B(h)-B(2*h)))/log(2);

wants to check B(2*h). The multiplication of a double with a logical returns a double. In case of the logical being false, the double will be zero. So, finally, you ask for B(0) and that does not exist in Matlab.

I guess there are quite some operations here which are not intended. Maybe you might want read about logical indexing as a keyword. Also, you can explain the purpose of your code and we might be able to hint you to working code.