Re: Slicing submatrices -- more graceful/ vectorized way?

[Jaroslav Hajek]

On Fri, Aug 20, 2010 at 7:03 PM, forkandwait <address@hidden> wrote:
> Hi all,
>
> Currently I have a loop that goes down a matrix and pulls out submatrices of 
> 36
>  rows (each submatrix is a county, and there are 39 counties, so there are 
> 1404
> rows total).
>
> Is there a more graceful way to handle this?  I always mess up my index
> calculations ("starti" and "endi" below) and it pisses me off.  Another
> constraint is that I would like to avoid 3-d matrices, though maybe they are 
> the
> best solution (I find that code that uses more than 2-d becomes brittle when
> other people try to maintain it -- hence the cellarray "x" to store each 
> county.)
>
>
> The code is something like this now:
>
> x={}
> cntyagesex = reshape(data(:,end), 39*36, []);
> for ii = 1:39
>  starti = (ii-1)*36 + 1;
>  endi = ii * 36;
>  x(ii) = cntyagesex(starti:endi,:);
> endfor
>
> I am hoping for a one-liner with kron() and reshape(), .... ;)
>

Hehe, yes, but I think the reshape would be 3D...
Face it: if you have a set of equal size, equal type matrices, 3D
array *is* the best solution. Still you can simplify your code, for
instance using the fact that ranges are expressions too:

cntyagesex = reshape(data(:,end), 39*36, []);
idx = 1:36;
for ii = 1:39
 idx += 36;
 x{ii} = cntyagesex(starti:endi,:);
endfor



-- 
RNDr. Jaroslav Hajek, PhD
computing expert & GNU Octave developer
Aeronautical Research and Test Institute (VZLU)
Prague, Czech Republic
url: www.highegg.matfyz.cz