Thread Subject:

problem with a matrix

I have a matrix (i,2) eg
table=1 2
         1 4
         1 5
         2 8
etc.
I want somehow to search the whole table and while the elements of the first column are the same, sum the elements of the second column till then and put them in a vector.
So here I should have vector=(11 8...)
Any ideas how to do this?I tried with "while" but it doesn't return the correct results.

"Konst " <konstance1@hotmail.com> wrote in message <jo3t63$3c8$1@newscl01ah.mathworks.com>...
> I have a matrix (i,2) eg
> table=1 2
> 1 4
> 1 5
> 2 8
> etc.
> I want somehow to search the whole table and while the elements of the first column are the same, sum the elements of the second column till then and put them in a vector.
> So here I should have vector=(11 8...)
> Any ideas how to do this?I tried with "while" but it doesn't return the correct results.

consolidator is on the file exchange. It can do a variety
of things beyond what you ask.

[uniquecol1,col2sums] = consolidator(table(:,1),table(:,2),@sum);

http://www.mathworks.com/matlabcentral/fileexchange/8354-consolidator

John

On 5/5/2012 1:57 PM, Konst wrote:
> I have a matrix (i,2) eg
> table=1 2
> 1 4
> 1 5
> 2 8
> etc.
> I want somehow to search the whole table and while the elements of the first
>column are the same, sum the elements of the second column till then and put them in a vector.
> So here I should have vector=(11 8...)
> Any ideas how to do this?I tried with "while" but it doesn't return the correct results.

may be one way

-------------------------
I = unique(A(:,1))
arrayfun( @(i) sum(A(A(:,1)==I(i),2)), 1:length(I))
---------------------------

ans =

     11 8

--Nasser

"Nasser M. Abbasi" <nma@12000.org> wrote in message <jo3u4p$s98$2@speranza.aioe.org>...
> On 5/5/2012 1:57 PM, Konst wrote:
> > I have a matrix (i,2) eg
> > table=1 2
> > 1 4
> > 1 5
> > 2 8
> > etc.
> > I want somehow to search the whole table and while the elements of the first
> >column are the same, sum the elements of the second column till then and put them in a vector.
> > So here I should have vector=(11 8...)
> > Any ideas how to do this?I tried with "while" but it doesn't return the correct results.
>
> may be one way
>
> -------------------------
> I = unique(A(:,1))
> arrayfun( @(i) sum(A(A(:,1)==I(i),2)), 1:length(I))
> ---------------------------
>
> ans =
>
> 11 8
>
> --Nasser
Thanks!It works fine!

"Nasser M. Abbasi" <nma@12000.org> wrote in message
> I = unique(A(:,1))
> arrayfun( @(i) sum(A(A(:,1)==I(i),2)), 1:length(I))
> ---------------------------

Arrayfun is like for-loop. Further more this code has N^2 complexity or array of size N (thus bad runtime for large array).

Consider using accumarray for large data:

[t1, ~, J] = unique(table(:,1));
[t1 accumarray(J,table(:,2))]

% Bruno

Another possibility,

vector=sparse(table(:,1),1,table(:,2));

"Matt J" wrote in message <jo5td3$753$1@newscl01ah.mathworks.com>...
> Another possibility,
>
> vector=sparse(table(:,1),1,table(:,2));

With few limitations:

1) something in table(:,1) is negative
2) something table(:,1) is not integer
3) One of the sum becomes zero

Bruno