How to concatenate 3D matrixes effectively/fast

Hi,

I have 5 matrixes, named M1,M2,M3,M4 and M5. They are all 48*1000*C, with C different in each case. I would like to concatenate M1 to M5, in dimension 3. I have written the following code:

ntables=40
nlines=1000
ncolumns1=10
ncolumns2=15
ncolumns3=20
ncolumns4=25
ncolumns5=30

M1=zeros(ntables, nlines, ncolumns1);
M2=zeros(ntables, nlines, ncolumns1);
M3=zeros(ntables, nlines, ncolumns2);
M4=zeros(ntables, nlines, ncolumns3);
M5=zeros(ntables, nlines, ncolumns4);


for i=1:nlines
    
    dataC1(:,i,:)=cat(3,M1(:,i,:),M2(:,i,:))
    dataC2(:,i,:)=cat(3,dataC1(:,i,:),M3(:,i,:))
    dataC3(:,i,:)=cat(3,dataC2(:,i,:),M4(:,i,:))
    dataC4(:,i,:)=cat(3,dataC3(:,i,:),M5(:,i,:))
    
end

The problem is that the code above is taking very long to run, like it is running for more than 1 week. Is there a faster and easier way to code this?

I thank you in advance.

Answer 1

[DGM]

... why not just concatenate them?

ntables=10
nlines=10
ncolumns1=10
ncolumns2=15
ncolumns3=20
ncolumns4=25
ncolumns5=30

M1=zeros(ntables, nlines, ncolumns1);
M2=zeros(ntables, nlines, ncolumns2);
M3=zeros(ntables, nlines, ncolumns3);
M4=zeros(ntables, nlines, ncolumns4);
M5=zeros(ntables, nlines, ncolumns5);

dataC = cat(3,M1,M2,M3,M4,M5);

If there's no need for M1, etc, then there's no need to concatenate at all.

ntables = 10
nlines = 10
ncolumns = [10 15 20 25 30];
dataC = zeros(ntables,nlines,sum(ncolumns));

Answer 2

[Hugo]

Hi,

Thank you for your reply. However, I am not trying to concatenate null matrixes. Instead, I am pre-allocating it. The matrixes M1....M5 are filled with values, not only 0.

[Walter Roberson]

ntables=10;
nlines=10;
ncolumns1=10;
ncolumns2=15;
ncolumns3=20;
ncolumns4=25;
ncolumns5=30;

%now for some example data, just to illustrate that it works.
%in your real code, you would create your real M1, M2, M3, M4, M5

M1=randi(10,ntables, nlines, ncolumns1);
M2=randi(10,ntables, nlines, ncolumns2);
M3=randi(10,ntables, nlines, ncolumns3);
M4=randi(10,ntables, nlines, ncolumns4);
M5=randi(10,ntables, nlines, ncolumns5);

%The matrixes M1....M5 are filled with values, not only 0.


dataC = cat(3,M1,M2,M3,M4,M5);

whos dataC

  Name        Size                Bytes  Class     Attributes

  dataC      10x10x100            80000  double              

[DGM]

Then the first example should suffice to concatenate them.