merged with head

matlab/util/group.m

@@ -0,0 +1,22 @@
+function [chunks, groupname, cuts] = group(M, group_on, aggregate)
+% Example Usage:
+% grps = group(raw.data, [1,2], [1,2,4]);
+% groups together (puts in the same cell) the rows of raw.data that have
+% the same values for columns [1,2]. And the data that is put in the cell
+% is the values in cols [1,2,4].
+
+
+%% This is a very powerful function to group large tables
+%  Let M be a large table with each row being a record
+%  then group on are the columns on which to group the records
+%  and aggregate are the columns to retain in the final grouping
+%  chunks will be a cell array where each cell contains all the
+%  records with groupname and cuts is the count of the items in that
+%  group.  
+[groupname, junk, ind] = unique(M(:,group_on), 'rows');
+clear('junk');
+[junk, oind] = sort(ind);
+clear('junk');
+cuts = hist(ind, min(ind):max(ind))';
+chunks = mat2cell(M(oind,aggregate), cuts, length(aggregate));
+end

src/graphlab/core.hpp

@@ -44,7 +44,7 @@ namespace graphlab {
      gl::core glcore;
      \endcode
    
-     The core contains the 
+cd     The core contains the 
    
      \li Data Graph: which represents the structured data dependencies.
      \li Engine: The computational structure which contains the