octave error: subscript indices must be either positive integers or logicals

Question

I'm trying to sum the product of an indexed vector and an indexed matrix like this:

k=[0:1:N-1]  
n=[0:1:N-1]  
x_n = sin(pi*n)  
N = size(x_n,2)  
_seqgen(x_n(n)*exp(k*n/N), n, 0..N-1)  

but I get the error:

error: subscript indices must be either positive integers or logicals  

What am I missing here?

EDIT: I just realized that I missed the _plus function to sum the generated sequences. It should look like this:

k=[0:1:N-1]  
n=[0:1:N-1]  
x_n = sin(pi*n)  
N = size(x_n,2)  
_plus(_seqgen(x_n(n)*exp(k*n/N), n, 0..N-1))  

I still get the same error though...

Solution 1

The error message explains what is wrong: you are trying to index an array with a number which is not a positive integer or logical. The only array indexing in your code is x_n(n). And sure enough, the line n=[0:1:N-1] demonstrates that the index n is not positive, since 0 is not a positive number. Lesson: MATLAB/Octave always index from 1. I do suggest you real some tutorials as this is fundamental stuff which you'll need to know.

Solution 2

Allow me to offer some critique of your code - since you admitted you were a newbie at this. First you create the vector

n = [0:1:N-1];

Which, incidentally, doesn't need the square brackets and could be written as

n = 0:N-1;

Then you generate a vector x_n which, for the values given, will be all zeros (sin(pi*n)==0 for integer values of n).

Next, you do something strange - you appear to be generating a symbolic sequence, looping a variable n which looks a lot like the array n you defined earlier. Not sure what to make of that - doesn't seem like a great idea. Notice that even @jazzbassrob was confused by this - the fact that you got a "can't index with zero" error wasn't because of the value of your vector n, but because you were looping from 0..N-1 in the _seqgen command (not the same thing, although it happens to be the same values).

In that _seqgen expression, I see exp(k*n/N) which works because in this context n is the variable being stepped through 0..N-1 - if Matlab was looking at the earlier definition of n, it would throw another error because of a dimension mismatch (since * is the matrix multiplication operator and expects the second dimension of the first element to be = the first dimension of the second element).

A more standard way to do what you are trying to do would be

mySum = sum(x_n.*exp(k.*n/N));

This does an element-by-element multiplication of the terms in x_n and the exp of a element-by-element product of k and n divided by N.

Note - whether this is actually "better" depends on what you want to do with the result (the above evaluates it).

Comments

[user7610]

In users @user2150186 defence, the error sounds as if there was something fundamentally wrong with the type (or class) of the indexing expression in itself. The message is too pompously general, given that the problem is one single incorrect value used as a subscript. "Zero is not a valid index" would be IMO much better.