concat/horzcat for cell with data cells of a*b*c sizes

Read variables from files

ncvars={'a','b'}; % variables need from files ncvar{1}=a, ncvars{2}=b
prjdir=[datadir 'mentioned path']; % path to directory
dinfo=dir(fullfile(prjdir,'*.nc')); %taking all files with extension .nc
num_files=length(dinfo); % number of files
E1=cell(num_files,1); % defining cell
E2=cell(num_files,1);
% taking a and b from files and putting in E1 and E2 respectively
for K = 1:num_files
    file=filenames{K};
    E1{K}=ncread(file, ncvars{1});
    E2{K}=ncread(file,ncvars{2});
end

Now the question is I want to join/concat/horzcat E1 and E2.

Where I have E1 and E2 of 2*1 cell. (2 is number of files)

When I had E1{1}=180*161 double and E1{2}= 180*146. I used folowing to horzcat:

Etemp=horzcat(E1{:},E2{:}); %Etemp is 180*614 where 614 because 2*(161+146)
E=Etemp(1:end) % E dimention 1*110520 double

Now I have E1{1}=180*161*253 double and E1{2}=180*146*127 double. Same for E2 too.

How can I use horzcat for E1 and E2 in this case??

So that I can have Etemp dimention someting like 180*100008 where because 2*(161*253+146*127)???

Answer 1

[Jan]

E1m = reshape(E1, 180, []);
E2m = reshape(E2, 180, []);
E   = cat(2, E1m, E2m);

cat(2, ...) is the same as horzcat().

[Ankitkumar Patel]

It is usefull some what. These gave divisible error.

Here E1 and E2 are 2*1 cell. So with some changes in these lines I able to create for E1 and E2 separately

E1m=reshape(E1{1},180,[]); % 180*40733
E11m=reshape(E1{2},180,[]); %180*18542
E1 =cat(2,E1m,E11m); %180*59275

Same way will get E2 and then after cat for E1 and E2 final result will be achived.

Is there any way where we can create loop for these??

[Jan]

Yes.

C = cell(1, 2);
for k = 1:2
    C{k} = reshape(E1{k}, 180, []);
end
E1M = cat(2, C{:});

A compact (but not faster) solution:

C = cellfun(@(x) reshape(x, 180, []), E1, 'UniformOutput', 0);
E1M = cat(2, C{:});

I use a new name instead of re-using E1. Letting a variable change its type from cell to double impedes the JIT acceleration. It is not a bug, but faster to avoid this.