How do I automate breakdown of an array into smaller arrays?

Hi everyone,

I have a number of nx3 arrays where n changes from data set to data set. I would like to take each data set (for example, randomly generated array 'r1' below, with 23 rows) and turn it into n-3 arrays, each with 4 rows (i.e. r1(1:4,:), r2(2:5,:),..., r3(20:23,:), and write each as a separate text file. I am stuck on the step of creating the n-3 (in this case 20, but again, n varies so I can't just assign a specific number of new arrays across all of my data sets) new arrays. When I run the following for loop:

>> for k=1:(size(r1,1)-3)
v=[r1(k:k+3,:)]
end

it outputs each of the arrays I want, but each time with the name 'v' so that when I export with writematrix(v), I can only export the last array output by the for loop (and the other 22 are lost). I would really rather avoid copy/pasting each 4x3 array into separate text files because I have a lot of data and it would take a long time. I also acknowledge that it's preferable to use efficient indexing over creation of dynamic variables as explained here https://www.mathworks.com/matlabcentral/answers/304528-tutorial-why-variables-should-not-be-named-dynamically-eval?s_tid=srchtitle. I'll keep working my way through the documentation but at this point I would welcome any guidance/syntax/functions/operations. Thanks! Also, example data set below:

>> rng('default')
>> r1=rand(23,3)

r1 =

    0.8147    0.9340    0.4456
    0.9058    0.6787    0.6463
    0.1270    0.7577    0.7094
    0.9134    0.7431    0.7547
    0.6324    0.3922    0.2760
    0.0975    0.6555    0.6797
    0.2785    0.1712    0.6551
    0.5469    0.7060    0.1626
    0.9575    0.0318    0.1190
    0.9649    0.2769    0.4984
    0.1576    0.0462    0.9597
    0.9706    0.0971    0.3404
    0.9572    0.8235    0.5853
    0.4854    0.6948    0.2238
    0.8003    0.3171    0.7513
    0.1419    0.9502    0.2551
    0.4218    0.0344    0.5060
    0.9157    0.4387    0.6991
    0.7922    0.3816    0.8909
    0.9595    0.7655    0.9593
    0.6557    0.7952    0.5472
    0.0357    0.1869    0.1386
    0.8491    0.4898    0.1493

[Stephen23]

"...it outputs each of the arrays I want, but each time with the name 'v' so that when I export with writematrix(v), I can only export the last array output by the for loop"

Have you considered the very obvious and simple solution of saving those matrices within the loop?

"I would really rather avoid copy/pasting each 4x3 array into separate text files because I have a lot of data and it would take a long time"

That would be a very poor approach, and should be avoided. Computers are only really good at one thing: simple operations repeated many times in a loop... so why do the computer's job for it?

"I would like to take each data set ... and write each as a separate text file."

What is stopping your from saving them as separated text files? So far you have not described any need to have them all (separately) stored in memory simultaneously.

[Dylan Barber]

Stephen, thanks for all of these comments.

I am pretty new here and don't currently know how to save the matrices within the loop. I'll look it up in the documentation but am also curious: what syntax would you use to accomplish this?

I would also love to save as separate text files, and do not need to save them simultaneously in the memory. I didn't mention this in the initial question, but I have been trying to figure out the syntax for that. So far I have been unsuccessful and would welcome any insight you might have on accomplishing it.

[Stephen23]

for k = ..
    v = ..
    f = sprintf('data_%d.txt',k);
    writematrix(v,f)
end

https://www.mathworks.com/help/matlab/import_export/process-a-sequence-of-files.html

[Dylan Barber]

Oh man. This is exactly what I needed. Thanks very much!

Answer 1

[David Fletcher]

for k=1:(size(r1,1)-3)
v(:,:,k)=[r1(k:k+3,:)]
end

All 4x3 matrices will be held in v indexable by the third dimension of v

[Dylan Barber]

Thank you David!