How can thousands of matrix lines be written into a multidimensional cell-array?

I have created a cell array as follows:

YY = cell(31,10);
for l = 1:numel(YY)
  YY{l} = zeros(1000,3);
end

YY has now 31 rows and 10 columns and each YY{i,j} has now 1000 rows and 3 columns.

The goal is now to fill the individual cells, with the elements of a matrix (which has some thousand lines and three columns).

I have tried the following to write the individual rows of the matrix into the respective cell:

% reading of simulated runoff (simulated)
S = 'runoff.txt'; % name of file (always the same)
D = 'C:\Users\heute\model\model_standalone\'; %pathname
d = dir(fullfile(D,'results*'));                         % file name is changing with consecuitive number
fmtS = ['%s',repmat('%f',1,6)];                            % Format 
opt = {'HeaderLines',1,'CollectOutput',true};              % headerline
L=length(d);                                               % number of sub-folders
C=zeros(L,3);                                          % preallocate

YY = cell(31,10); %create cell array
for l = 1:numel(YY) % create cells with a dimension of (1000,3)
  YY{l} = zeros(1000,3);
end

for k = 1:L  % loop to open files, make calculations and fill the cell arrays

    fid = fopen(fullfile(D,d(k).name,S),'rt');               % open the files with simulated data
    Z=textscan(fid,fmtS,opt{:});                             % read simulated data
    fclose(fid);                                             % close file
    dtS=Z{:,1};                                              % timestamp simulated (datetime)
    Qs=Z{:,2}(:,6);                                          % variable with simulated runoff

    % Define timesteps, which are used for analysis
    yr1=2014; yr2=2016;                                      
    ix=iswithin(dto,datenum(yr1,05,01),datenum(yr1,10,01));  % first year
    for yr=yr1+1:yr2                                         % subsequent years
      ix=ix | iswithin(dto,datenum(yr,05,01),datenum(yr,10,01));
    end
    f_1k = 1-cov(Qs(ix)-Qo(ix))/var(Qo(ix));                 % NSE
    f_2k = rms(Qs(ix)-Qo(ix));                               % RMSE
    f_3k = abs(mean(Qs(ix)-Qo(ix)));                         % BIAS

    C(k,:) = [f_1k, f_2k, f_3k];  

    % write matrix C into the cell array
    YY{k} = C;

end

With this variant, all individual cells are filled with the same values.

Another variant with a for - loop writes only one row in each cell. In this case four cells are filled with only one row.

for ii=1:4
   YY(ii) = {C(ii, :)};
end

How can you solve the problem, that the first 1000 values are written in the first cell, the second 1000 values in the second cell etc?

[Glazio]

Have still found no working solution.

Does anyone have an idea how to solve this problem?

Answer 1

[Walter Roberson]

Write them into a large numeric array, possibly a 4 dimension one. mat2cell() afterwards.

[Glazio]

@ Walter Roberson: Thanks for your help. If I use the following code:

YY = mat2cell(C,[31,1000], [10,3]);

this error message occurs:

Error using mat2cell (line 107) Input arguments, D1 through D2, must sum to each dimension of the input matrix size, [310000 3].'

If I am not wrong, first the line numbers and then the column numbers are defined?

[Image Analyst]

I'm totally confused as to why you want a cell array anyway. Why not just simply use a double array?

[Walter Roberson]

YY = mat2cell(C, 1000 * ones(1,31), 3 * ones(1,10));

[Glazio]

@ Walter Roberson,

thanks for your answer, but the same error message occurs:

Error using mat2cell (line 107) Input arguments, D1 through D2, must sum to each dimension of the input matrix size, [310000 3].'

Another error message in this context is:

Error using mat2cell (line 97) Input arguments, D1 through D2, should be vectors.

I have no idea where the errors might originate.

[Glazio]

@ Image Analyst:

Further evaluation steps and scripts require a cell array as input.

[Walter Roberson]

All of the material before has said that you have 31 rows of cells and each cell entry has 1000 numeric rows. That would be 31 * 1000 = 31000 total rows. But the error message is talking about 310000 total rows, which is 10 times larger. You need to figure out where the factor of 10 larger is coming from in your C, and perhaps decide if you want 310 rows of cells of 1000 rows each or if you want 31 rows of cells of 10000 rows each.

number_of_rows_of_cells = 31;
number_of_rows_in_each_cell = 1000;
number_of_columns_of_cells = 10;
number_of_columns_in_each_cell = 3;

YY = mat2cell(C, number_of_rows_in_each_cell * ones(1,number_of_rows_of_cells), number_of_columns_in_each_cell * ones(1,number_of_columns_of_cells));

You might notice that [310000 3] is simultaneously off by a factor of 10 in the number of columns: to have 10 columns of cells each with 3 columns, you need to have started with 30 columns, not with 3. This factor of 10 is the same as the factor you are off on the number of rows, suggesting that you have created rows where you wanted to create columns.

Depending on how you did the construction you just might be lucky enough to be able to use

YY = mat2cell( reshape(C, [], 30), 1000 * ones(1,31), 3 * ones(1,10));

but it is certainly possible that you will need to do a reshape() then a permute() then another reshape() in order to get the data into the proper locations to use mat2cell()

[Glazio]

@ Walter Roberson: It now seems that the desired results are provided, Thanks for your help.