prmem.m

The pattern recognition matlab toolbox

%PRMEM Size of memory and loops for intermediate results
% 
%   [LOOPS,ROWS,LAST] = PRMEM(M,K)
% 
% Assume that an array of the size [M x K] has to be computed. The 
% numbers of LOOPS and ROWS are determined which are needed such that 
% ROWS*K < GLOBALPRMEMORY (a global variable that is initialized in this 
% routine, if necessary). The final number of rows for the last loop 
% is returned in LAST. 
%
% EXAMPLES
% [M,K] = size(A);
% [LOOPS,ROWS,LAST] = prmem(M,K);
% if (LOOPS == 1)
%  RESULT = < compute the result based on A >
% else
%   RESULT = 0;
%   for J =1:LOOPS
%     if (J == LOOPS), N = LAST; else N = ROWS; end
%     NN = (J-1)*ROWS;
%     RESULT = RESULT + < compute a partial result based on A(NN+1:NN+N,:) >
%	  end
% end