Saving, extracting relevant data from cell arrays in multiple loops in MATLAB

Question

sorry for asking may be this is simple question but I am stuck in it. It’s related to cell arrays.

I have delay, offset, and threshold.

 delay = [0.01:0.01:0.03];
 offset = [0.02:0.01:0.04];
 threshold = [0.4:0.1:0.9];       
 limit_for_idx  = [0.4:0.1:0.9]; 
 limit = [0.4:0.1:0.9];

I have tried to provide an example to ask only a part where I have some issue.

In first part of loop I have received complete values for Delay, Offset, J, r, threshold as shown in figure for all loops.

delay =0.0300 offset=0.0400 J=16 25 24 25 r = 24 21 46 18 threshold = 0.4:0.9

Then I filtered Minimum value of J and maximum value of r and threshold values which corresponds to min J and max r. I have received these values as shown in figure.

   [min_J,min_J_loc] = min(J(:))
   [max_r,max_r_loc] = max(r(:))
   thresh_min_J = threshold(min_J_loc);
   thresh_max_r = threshold(max_r_loc);

For the one case I have marked with red but I will receive these combinations for all loops like shown in whole picture.

Questions:

• I want to find the combination of delay, offset, threshold which contains minimum J and maximum r as shown in figure from all the combinations I need I need the combination of delay ,offset, threshold which contain minimum J and maximum r . I have saved values in the cell array as I am confused that how to get the combination out of cell array.

• Is there any good approach to save the values or may be structures if yes can someone explain how just small hint will also work.

Code :

 delay = [0.01:0.01:0.03];
 offset = [0.02:0.01:0.04];
 threshold = [0.4:0.1:0.9];       
 limit_for_idx  = [0.4:0.1:0.9]; 
 limit = [0.4:0.1:0.9];         


 J=0;
 T = 1;
 b=1;
 K=1;
 for H = 1:numel(delay)
     for G = 1:numel(offset)
        for R = 1:numel(threshold);  

          J = randi([10 25],1,4);    
           r = randi([10 50],1,4);
        end
   [min_J,min_J_loc] = min(J(:))
   [max_r,max_r_loc] = max(r(:))
   thresh_min_J = threshold(min_J_loc);
   thresh_max_r = threshold(max_r_loc);
       out{K,:} = [ delay(H) offset(G)  J  r threshold];
       output{T,:} = [delay(H) offset(G)  min_J  max_r  thresh_min_J  thresh_max_r];
     K=K+1;
     T = T+1;
     end

 end


  for X = 1:numel(out)
      disp(' delay ,   offset(G) ,  J,   r ,  threshold  ') 
      Q = out{X};
      disp(Q)
  end

   for X = 1:numel(output)
        disp(' delay ,   offset(G) ,  min_J,   max_r ,   thresh_min_J  thresh_max_r ')
      Z =  output{X};
      disp(Z)

   end 

Answer 1

Check out this modified version of your example:

function varargout = q47452082
delay = (0.01:0.01:0.03);
offset = (0.02:0.01:0.04);
threshold = (0.4:0.1:0.9);
nD = numel(delay);
nO = numel(offset);
J = randi([10 25],1,4,nD,nO);
r = randi([10 50],1,4,nD,nO);
%% Preallocate
%{
The simplest way to do it:
out    = NaN(nD*nO,3*1+2*4);
output = NaN(nD*nO,6);
%}
out = struct('delay',[],'offset',[],'J',[],'r',[],'threshold',[]);
out = repmat(out,nD*nO,1);
output = struct('delay',[],'offset',[],'min_J',[],'max_r',[],...
  'thresh_min_J',[],'thresh_max_r',[]);
output = repmat(output,nD*nO,1);
fn{2} = fieldnames(output);
fn{1} = fieldnames(out);
%% Populate the data structures:
K = 1;
for H = 1:numel(delay)
  for G = 1:numel(offset)
    [min_J,min_J_loc] = min(J(:,:,H,G));
    [max_r,max_r_loc] = max(r(:,:,H,G));
    thresh_min_J = threshold(min_J_loc);
    thresh_max_r = threshold(max_r_loc);
    data = {delay(H),offset(G),J(:,:,H,G),r(:,:,H,G),threshold};
    for indF = 1:numel(fn{1})
      out(K).(fn{1}{indF}) = data{indF};
    end
    data = {delay(H), offset(G), min_J, max_r, thresh_min_J, thresh_max_r};
    for indF = 1:numel(fn{2})
      output(K).(fn{2}{indF}) = data{indF};
    end
    K = K+1;
  end
end

if nargout == 0 % if no outputs requested, print
  if ~verLessThan('matlab','8.2') % tables exist in MATLAB R2013b or newer
    disp(struct2table(out));
    disp(struct2table(output));
  else
    for X = 1:numel(out)
      Q = out(X);
      disp(Q)
    end
    for X = 1:numel(output)
      Z = output(X);
      disp(Z)
    end
  end
else % otherwise output the desired data:
  % OPTION #1: separate variables
  % You should call the function like so: [min_J_cases,max_r_cases] = q47452082();
  varargout{1} = output([output.min_J] == min([output.min_J]));
  varargout{2} = output([output.max_r] == max([output.max_r]));
  % OPTION #2: 1 output, 2x1 cell
  %{
  varargout = {output([output.min_J] == min([output.min_J]));...
               output([output.max_r] == max([output.max_r]))};
  %}
  % OPTION #3: 1 output, 2x1 struct
  %{
  varargout = {[output([output.min_J] == min([output.min_J]));...
                output([output.max_r] == max([output.max_r]))]};
  %}
end

Several things you should note:

• I removed some unused variables.
• I preallocated out and output as struct arrays.
• I removed the loop around the J,r randomization (since it does nothing) and moved the random number generation to before the loop (no reason to do it in every iteration when it can be done all at once at the beginning).
• I perform assignment using dynamic field references to both struct arrays.
• I added a demonstration of how to run different code depending on different MATLAB version (and hence, different supported functionality) - see printing at the end.
• The function only prints things if you don't request outputs.
• There is likely room for further vectorization of your code (i.e. getting rid of loops).