Using a while loop on each column in an array?

Hi all,

I have a script that performs an action on each column of an array at once. I would like to set up a loop where I do this multiple times. Each time I will be taking the mean of each column and comparing it to another value. I would like to know at which iteration each column's mean reaches that value, although each column will do this at different times. With each iteration the columns will get 1 longer, thus making the mean more accurate.

I don't want my script to stop when only one of the columns reaches this point, but I also don't want it to stop and only record the iteration when all of them have reached this value.

For instance:

 B = 4;

 for i = 1:4
    if i == 1
        A = [1,2,3;1,2,3;1,2,3;4,4,4];
    elseif i == 2
        A = [1,2,3;1,2,3;1,2,3;4,4,4;4,4,4];
    elseif i == 3
        A = [1,2,3;1,2,3;1,2,3;4,4,4;4,4,4;4,4,4];
    elseif i == 4
            A = [1,2,3;1,2,3;1,2,3;4,4,4;4,4,4;4,4,4;4,4,4];
    end 
  j = mean(A)/B;            
 end

If I want j > .65, I would like a vector that tells me column one reached this at i == 4, column 2 reached this at i == 2 and column 3 reached this at i == 1. If I didn't use a while loop I do have a fixed number of iterations I could do, I would just like to avoid doing them all if I can help it. But maybe I could do that and then go back with cumsum? Any help would be appreciated, thank you!

[CS Researcher]

First of all mean(A)/B will be a vector and not a scalar value. Also, I am not sure I completely understood your problem. Can you elaborate your example further?

[Tyler]

Yes, sorry for the confusion. You're right, j will be a vector with the mean of each column. I would like to know when each column of j is greater than .65 for example. With each loop my output array A gets bigger(in this example I entered in A manually, but in my script it's more complicated), and my j vector changes each time. However, my problem is that I would like to do this for more than 4 iterations- more like 260 at the most. But I would like it to stop when all the columns have reached that value. I would also like to know when each of the other columns reached above .65.

So for this example,

 j(1) = [.4375 .6250 .8125]
 j(2) = [.5500 .7000 .8500]
 j(3) = [.6250 .7500 .8750]
 j(4) = [.6786 .7857 .8929]

I would like an output vector with values of:

 [4 2 1]

where column 1 was > .65 when i == 4, column 2 reached > .65 when i == 2, and column 3 was > .65 when i == 1.

Also, if I didn't know how many iterations it would take, I would like to set up the loop with

 for i = 1:100
   blah blah blah
end

But then I would like it to stop at i == 4 and not continue. Does this make it clearer? Sorry it's taking so long and proving difficult to explain my problem- thanks for your time!

Answer 1

[Guillaume]

I'm not entirely clear on everything particularly on the stop condition for your loop.

You can indeed calculate the result a posteriori:

A = [1,2,3;1,2,3;1,2,3;4,4,4;4,4,4;4,4,4;4,4,4];
cummean = bsxfun(@rdivide, cumsum(A), (1:size(A, 1))');
%we can use max to find the first row of each column above a threshold or use a loop
[~, firstrow] = max(cummean / 4 > 0.65, [], 1)
firstrow = firstrow - 4

If you want to do the tracking in the loop, it's also trivial to do:

fullA = [1,2,3;1,2,3;1,2,3;4,4,4;4,4,4;4,4,4;4,4,4];
firstrow = zeros(1, size(A, 2)); 
for iter = 1 : size(fullA, 1)
  A = fullA(1:iter, :);  
  abovethreshold = mean(A) / 4 > 0.65;
  firstrow(abovethreshold & ~firstrow) = iter;
  if all(firstrow)
     break; %is this your stop condition?
  end    
end