Random but unique values in matrix
12 views (last 30 days)
Show older comments
Please help me! I want to generate random but unique row as given in code. I want generate complete 100*13 matrix. When this code run it stuck after around 29th iterations. Please help me and give suggestions. If my question is not understandable please let me know. I will explain more in detail.
j=1;
pop_size=100;
V=13;
while j:pop_size
y=randi([0 1],j,V);
y(:,2)=~y(:,1);
y(:,3)=~y(:,4);
y(:,5)=~y(:,6);
y(:,12)=~y(:,13);
y = unique(y, 'rows');
j=size(y,1)+1;
end
10 Comments
Steven Lord
on 29 Aug 2016
So generate a random 100-by-8 matrix. Check if all its rows are unique using the unique function with the 'rows' flag. Once you have a suitable matrix, expand it to 12 columns. [Once columns 1, 3, 5, and 12 are fixed they determine columns 2, 4, 6, and 13.]
Accepted Answer
dpb
on 29 Aug 2016
>> z=randi([0 1],100,13); % Generate trial array
>> [u,ia,ib]=unique(z,'rows','stable'); % check for uniqueness between rows
>> length(ia) % dang!, one short...
ans =
99
>> find(ismember(z,z(setdiff([1:100],ia),:),'rows')) % locate which are the offending rows
ans =
34
92
>>
>> [z(34,:);z(92,:)] % indeed, they are identical...
ans =
0 0 1 0 0 1 1 0 1 0 1 0 0
0 0 1 0 0 1 1 0 1 0 1 0 0
>>
OK, NOW you can begin a loop looking to generate a new row that is different than z(34,:) and also not duplicate another already...
2 Comments
Thorsten
on 29 Aug 2016
Or you generate a trial array with many more rows than actually needed and hope that you'll end up with more than 100 unique rows. See my answer below.
BTW: You forgot the ensure the four constraints
y(:,2)=~y(:,1);
y(:,3)=~y(:,4);
y(:,5)=~y(:,6);
y(:,12)=~y(:,13);
More Answers (3)
Stephen23
on 29 Aug 2016
Edited: Stephen23
on 29 Aug 2016
A complete working example:
X = randi([0,1],100,9);
chk = true;
while chk
[~,idx] = unique(X,'rows');
idy = setdiff(1:100,idx);
X(idy,:) = randi([0,1],numel(idy),9);
chk = numel(idy)>0;
end
Y = X(:,[1,1,2,2,3,3,4:end,end]);
idz = [1,3,5,12];
Y(:,idz) = ~Y(:,1+idz);
and tested:
>> size(unique(Y,'rows'))
ans =
100 13
>> Y(:,1:2) % check that two columns are always different
ans =
0 1
1 0
1 0
1 0
0 1
1 0
1 0
1 0
0 1
1 0
1 0
etc
>> Y(:,12:13)
ans =
1 0
1 0
0 1
1 0
1 0
0 1
1 0
1 0
1 0
0 1
0 1
0 1
1 0
0 1
etc
3 Comments
antlhem
on 28 Apr 2019
Edited: antlhem
on 28 Apr 2019
Hi Stephen Cobeldick, how would you suggest me to use this method in order to create a bigger array (1500x1500). With all the rows unique among each other?
Ori Golani
on 9 Oct 2018
Hi all, I know that this is an old thread, but I found an efficient non-iterative solution for this.
Complete working example:
I= 2; % maximal integer in the matrix (values will be [0...I-1])
M= 3; % Matrix rows
N= 10; % matrix columns
word_indices= randperm(I^N,M);
matrix_in_char= dec2base(word_indices,I);
random_matrix= reshape(base2dec(matrix_in_char(:),I),M,N);
Tested:
random_matrix =
1 1 1 0 1 1 1 0 1 0
0 0 1 1 0 1 0 0 0 1
0 0 1 0 1 0 1 1 1 0
2 Comments
antlhem
on 28 Apr 2019
Edited: antlhem
on 28 Apr 2019
Hi Ori Golani, do you know how to create a bigger matrix with this method? I tested with M=N=1500, but is not passing this test. There is an error from MATLAB saying:
Error using randperm
N must be less than 2^53.
Any suggestions, to be able to create bigger random-unique matrix?
Thorsten
on 29 Aug 2016
Edited: Thorsten
on 29 Aug 2016
The trick is to get more rows then actually needed, like 2*pop_size. Then you'll get your y almost always in the first trial:
pop_size=100;
V=13;
trials = 1;
while 1
y = randi([0 1], 2*pop_size, V);
y(:,2)= 1- y(:,1);
y(:,3)= 1- y(:,4);
y(:,5)= 1- y(:,6);
y(:,12)= 1- y(:,13);
[yu, xi] = unique(y, 'rows');
if size(yu, 1) >= pop_size
y = y(xi(1:pop_size),:);
break;
else
trials = trials + 1;
end
end
% check
size(unique(y, 'rows'), 1)
2 Comments
See Also
Categories
Find more on Creating and Concatenating Matrices in Help Center and File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!