maxminco.m

FISMAT accommodates different arithmetic operators, fuzzification and defuzzification algorithm, imp

function B=maxminco(A,M)

% B=maxminco(A,M)%
% max-min composition [Klir 1988]
%
% Fuzzy vector-matrix multiplication (Kosko (1992) pp.307) :
%   For row fit vectors A and B and fuzzy n-by-p matrix M.
%   The "recalled" component b(j) is computed by taking the fuzzy inner
%   product of fit vector A with the jth collumn of M.
%   The fit vector A is composed with the fuzzy matrix M to get the 
%   "recalled" vector B.
%   If we somehow encoded (A,B) in the FAM matrix M, we would say that 
%   the FAM system exhibits perfect recall in the forward direction.
%
% Possibilistic matrix product: 
%   (use like stochastic transitional probability matrix for discrete 
%    Markov processes)
%   The intersection/conjunction in prob. theory realized as a product of
%   two probabilities is equivalent to the minimum of two possibilities.
%   The union/disjunction in prob. theory realized as an addition of two
%   probabilities is equivalent the maximum of two possibilities. Thus
%   for the conventional transition-matrix product multiplication, in
%   possibility theory the product is replaced by pairwise minima and 
%   the addtion is replaced by maximum operation. Is A only a row vector,
%   the result is identical to the description above.
%
% FSTB - Fuzzy Systems Toolbox for MATLAB
% Copyright (c) 1993-1996 by Olaf Wolkenhauer
% Control Systems Centre at UMIST
% Manchester M60 1QD, UK
%
% 20-Feb-1994
% Last change: 15 Sep 1995

% Code used until 15 Sep 95:
%[n,p]=size(M); B=zeros(n,p);
%for i=1:n,
%  B(i,:)=min((ones(1,p).*A(i)),M(i,:));
%end;
%B=max(B);

[r1,c1]=size(A); [r2,c2]=size(M);
B=zeros(r1,c2);
for i=1:r1, % for each row vector in A
   B(i,:)=max(min(A(i,:)'*ones(1,c2),M));
end;