How to combine random size arrays in one matrix - in a loop?

Hey, I've been able to find a lot of solutions for my matlab program on this site but the next problem still goes unanswered.
I have a loop in which every run a vector is created with a random size. After the loop is finished I want all these vectors presented in a matrix with each vector as a column. How can I do this? The shorter vectors may either be filled with NaN's or zero's.
In essence the program looks like this:
for i=1:1:10;
l=ceil(rand(1)*10);
vector=rand(l,1);
end
matrix=[vector1 vector2 ... vector10 ];
Thanks in advance for helping me out.

 Accepted Answer

Hello Jeroen,
  • Generate the vectors - I'll call them b - in your for-loop and keep track of the longest vector that is generated in the loop. Store the vectors as a matrix: b(:,ii).
  • Once done, initialize a 0-matrix A using zeros(M,N) where M is equal to the size of the longest vector and N is the total number of vectors.
  • Lastly run another for-loop to paste into the matrix the individual vectors using something like:
A(:,ii) = [b(:,ii); zeros(length(A(1,:)) - length(b(:,ii)),1)]
where ii is the running index of the loop.
You can also get this done using only one loop, of course, by simply concatenating vectors that have been "filled up" with the adequate number of 0s, or filling up the existing matrix before concatenating.

2 Comments

Run the code below. Not optimized but it does what you are looking for:
A = [0];
for ii = 1:10
b = ones(randi([1,10],1,1), 1); % here you put your random-sized vector
if (size(A(:,1)) == 1)
A = b;
else
if (length(b) < length(A(:,1)))
A = [A [b; zeros(length(A(:,1)) - length(b),1)]];
end
if (length(b) == length(A(:,1)))
A = [A b];
end
if (length(b) > length(A(:,1)))
A = [[A; zeros(length(b) - length(A(:,1)), length(A(1,:)))] b];
end
end
end
display(A);
it works in between the program but this whole code does not work if I created as function

Sign in to comment.

More Answers (3)

You have several options how to store vectors with different lengths.
1) use cell arrays
C = cell(k,1) ;
for k=1:10,
L = ceil(10*rand) ;
C{k} = rand(L,1) ;
end
2) subsequently you can concatenate these cells into a single array, in which shorter vectors are padded with a chosen value. I have submitted the function PADCAT to the file exchange, which makes this a trivial conversion:
[M, tf] = padcat(C{:}) % pad with NaNs by default
M(~tf) = 0 % change NaNs to zeros
how about:
matrix = zeros(10,10);
flag = 1;
for i = 1:10
m = randi(10);
matrix(1:m,flag:flag+m-1) = rand(m);
flag = flag + m;
end
for jj = 1:10
v = randi(randi(234),randi(12),1);
if jj==1
out = v;
else
n = [size(out,1),numel(v)];
out(n(1)+1:n(2),1:jj-1) = nan;
out(1:max(n),jj) = nan;
out(1:n(2),jj) = v;
end
end

2 Comments

All great answers The simpler the better!!
Thanks. It works fine.
it works in between the program but this whole code does not work if I created as function

Sign in to comment.

Categories

Asked:

on 15 Jan 2014

Edited:

on 3 May 2018

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!