Search All of the Math Forum:

Views expressed in these public forums are not endorsed by NCTM or The Math Forum.

Notice: We are no longer accepting new posts, but the forums will continue to be readable.

Topic: Subscripted dimension mismatch
Replies: 2   Last Post: Jan 14, 2013 10:23 AM

 Messages: [ Previous | Next ]
 Steven Lord Posts: 18,038 Registered: 12/7/04
Re: Subscripted dimension mismatch
Posted: Jan 14, 2013 10:23 AM

"Jacob Moses" <jacobfmoses@gmail.com> wrote in message
news:kcs0nh\$aav\$1@newscl01ah.mathworks.com...
> Hello,
> I am attempting to separate a large matrix into two parts for kernel
> formation for clustering with a kernel k-means algorithm. While I would
> ordinarily jus reshape it into a third dimension, this option is
> impossible given that the number of columns in the matrix in question is
> odd (24021). I have been unable to perform any of the things that I have
> seen before, such as indexing. Below is the method I tried, which returned
> the above error message.
>
> EDU>> for n=1:24041
> if n<=12021
> split{1}(n)=test(:,n);

Unless test is a row vector, this won't work. The expression on the left of
the equals sign represents a scalar (a 1-by-1 chunk of split{1}) while the
expression on the right of the equals sign represents a size(test, 1)-by-1
chunk of data. The latter is too big to fit into the former unless as I said
size(test, 1) is 1.

> elseif n>12021
> split{2}(n-12021)=test(:,n)

Ditto.

> end
> end
> Subscripted assignment dimension mismatch.
>
> If anyone can tell me what is going on, I would greatly appreciate it.

There's no need for a loop here. Either:

1) If test is of odd length, pad it with a NaN then RESHAPE it. Otherwise
just RESHAPE it.
2) split{1} = test(:, 1:12021); split{2} = test(:, 12022:end);

If test may vary in size, you could use floor(end/2) in the second
expression to generalize it.

--
Steve Lord
slord@mathworks.com