loop runs infinitely,

% member ship value taken by zeros initilization
U=zeros(2,3);
% **********************************************

%    each row is randomly initialize and 
%                  must be 1 [ i.e., summation of complete row value]
% **********************************************
for i=1:2
%     for each i is 1 to 2, iteratively increased row
    while(1) 
        total=0;
        for j=1:3            
            U(i,j)=rand();
            % generaly random value is less than one
            total=total+U(i,j);
            % each row, every colomn value is added to variable 'total'
        end
        if (total == 1)
            % check value is equal to one or not,example total=1.1235
            % then total considers as 1 but my point is exact total
            % varible must be 'one'
            break;
        end
    end       
end 
disp(U);

loop runs infinite so that please describe how to stop it

[Adam]

while(1)

defines an infinite loop. It's always dangerous to do if you cannot be 100% certain that within the loop you have a guaranteed termination criterion.

[pavan kumar teegala]

my point, a row array is random generated value with keyword--> rand()

                 but all row values must be equal to 'one'.
ex:  
              col 1   col 2   col 3
     row 1  : 0.1254  0.6412  0.23334
     row 2  : 0.525   0.3500  0.125

     row 1  : (0.1254+0.6412+0.23334)=1
     row 2  : (0.525+0.3500+0.125)=1

[Adam]

Yes, but it doesn't take much to realise that will pretty much never happen, as Image Analyst points out. What do you think the percentage chance is of 3 totally random numbers of ~15 decimal places summing to exactly 1?, even before you take into account floating point inaccuracies within that sum even if they theoretically did sum to 1.

[pavan kumar teegala]

Thank You @Adam, I tried many times to do theoretically and finally I got practical implementation on the code. just while condition with flag variable assume as true, if inner condition fails then flag variable assume as false.......

function [U]=memv(mrow,k) % memv- fuzzy membership value [0,1] and mrow --> max rows of data % k --> no of clusters U=zeros(mrow,k);

for r=1:mrow % for each i is 1 to 2, iteratively increased row flag=true; while(flag) total=0; for cl=1:k

                U(r,cl)=rand();
%                 generaly random value is less than one
                total=total+U(r,cl);
%                 each row, every colomn value is added to variable 'total'
             end
             smallNeglate=0.00001;
            if abs(total -1) <smallNeglate
%                 check total value is equal to one or not,example
%                   total=1.1235 in matlab then total considers as 1 
%                    but my point is exact total varible must be 'one'
                  flag=false;
            end
     end

end % disp(U);

[Image Analyst]

I don't see any questions so I assume you have it solved now. However there is the implied question of "How do I format my code so it is readable?" and here is the answer to that question: http://www.mathworks.com/matlabcentral/answers/13205#answer_18099

Answer 1

[Image Analyst]

Since total is random numbers, it will never equal exactly 1, so see if it exceeds 1 instead

if total >= 1

Of check within a tolerance

if abs(total-1) < someSmallNumber

Please read the FAQ for more info: http://matlab.wikia.com/wiki/FAQ#Why_is_0.3_-_0.2_-_0.1_.28or_similar.29_not_equal_to_zero.3F

Answer 2

[Roger Stafford]

Oops! I've done something wrong and seemingly erased both my answer and Jan Simon's. I don't know how to correct it, but a thousand pardons Jan!

[Image Analyst]

The way to correct it is to use the Lazarus browser plugin. It's save me numerous times! Highly recommended!

[Jan]

@Roger: Never mind. I have a bunch of answers here in my drawer and simply choose another one. :-)

Answer 3

[Jan]

You can either normalize the data:

U = rand(2, 3);
U = bsxfun(@rdivide, U, sum(U, 2));

Or use Roger's smart function FEX: random-vectors-with-fixed-sum