Thread Subject:

Concatenate vectors of different lengths (auto-pad?)

I wonder if anyone has a nice way of automatically padding vectors (or
matrices) with 0 so they can be concatenated.

For example, suppose I have vectors
A = [1 2 3]
B = [3 0 4 0 5]
C = [2 1]

I would like to have a function supervertcat(A,B,C) that outputs the matrix
[1 2 3 0 0
3 0 4 0 5
2 1 0 0 0]

I would also like this to work for matrices:
A = [1,2;...
     3,4]
B = [1,2,3]
C = supervertcat(A,B) =
[1,2,0
3,4,0
1,2,3]

Does anyone have a nice way of doing this? I have seen the file exchange
submissions PADADD, PADCAT, and CATPAD, which seem fine, but are a bit
more heavyweight than I am seeking. I suspect there is a cute 2- or
3-liner out there somewhere.

Alan Weiss
MATLAB mathematical toolbox documentation

Alan_Weiss <aweiss@mathworks.com> wrote in message <jrakm2$ncp$1@newscl01ah.mathworks.com>...
> I wonder if anyone has a nice way of automatically padding vectors (or
> matrices) with 0 so they can be concatenated.
>
> For example, suppose I have vectors
> A = [1 2 3]
> B = [3 0 4 0 5]
> C = [2 1]
>
> I would like to have a function supervertcat(A,B,C) that outputs the matrix
> [1 2 3 0 0
> 3 0 4 0 5
> 2 1 0 0 0]
>
> I would also like this to work for matrices:
> A = [1,2;...
> 3,4]
> B = [1,2,3]
> C = supervertcat(A,B) =
> [1,2,0
> 3,4,0
> 1,2,3]
>
> Does anyone have a nice way of doing this? I have seen the file exchange
> submissions PADADD, PADCAT, and CATPAD, which seem fine, but are a bit
> more heavyweight than I am seeking. I suspect there is a cute 2- or
> 3-liner out there somewhere.
>

4 lines, and not very cute, but anyway here is my proposal:

function C = supervertcat(varargin)
% C = supervertcat(A1, A2, ...)

[i j v] = cellfun(@find, varargin, 'uni',0);
r = [0 cumsum(cellfun('size',varargin,1))];
[i j v] = arrayfun(@(k) deal(r(k)+i{k}(:),j{k}(:),v{k}(:)), 1:nargin,'uni',0);
C = accumarray([cat(1,i{:}) cat(1,j{:})], cat(1,v{:}));

end

If it was my code, I probably do something like this:

function C = supervertcat(varargin)
% C = supervertcat(A1, A2, ...)

ncol = cellfun('size',varargin,2);
maxcol = max(ncol);
for k = 1:nargin
    if ncol(k) < maxcol
        varargin{k}(end,maxcol) = 0;
    end
end
C = cat(1, varargin{:});

end

On 6/13/2012 3:44 PM, Bruno Luong wrote:
> If it was my code, I probably do something like this:
>
> function C = supervertcat(varargin)
> % C = supervertcat(A1, A2, ...)
>
> ncol = cellfun('size',varargin,2);
> maxcol = max(ncol);
> for k = 1:nargin
> if ncol(k) < maxcol
> varargin{k}(end,maxcol) = 0;
> end
> end
> C = cat(1, varargin{:});
>
> end
Thanks, Bruno, this is pretty clear.

Alan Weiss
MATLAB mathematical toolbox documentation