How to save a variable within a parfor loop?

Hello saviors!

so, I need help: I'm trying to run a parfor loop nested with a forloop, and save the outcome in a matrix array named x_mean that was previously defined , I'm doing this:

muvec= cat(2,fliplr(180:-deltin:inferior),180+deltin:deltin:superior);

delvec= muvec - m(1);            %vector array with delta m2
l_vec= length (muvec); 
rep=10;                          %number of replicas

x_mean=zeros (rep,l_vec,n);      %matrix array to save the output (x_mean definition)

tspan = y*T; 
x_ini=zeros (rep,n);             %vector array to save all the generated random x0%  (do not pay attention to this,
%  this concern to another part of the code)

%here I start de nested parfor:
parfor fu=1:rep
    %This dummys variables are a duplicate of the above and are used to decrease run time
    dummy_x0=x_ini;             % x_ini dummy
    dummy_xmean=x_mean;         % x_mean dummy ATENTION!
    dummy_muvec= muvec;         % muvec dummy
    dummy_m=m                   % m dummy
   %this parfor calls each x0 in the array x_ini (using its dummy) to be used in the ode45 
    x0=dummy_x0(fu,:);
    
    %this for loop calls each mu value in the array muvec (using its dummy)
    %to be used in the ode 45
      for ind=1:l_vec
          dummy_m(2)=dummy_muvec(:,ind);
         
                  
 
 [t, x] = ode45(@odesystem, [0, tspan], x0, [], ...
               n, K, N, S, A, T, b, s, p, f, m, w, a, d);
           
  t_uso = find(t>730);                   % find the indices where t>2 years
  x_last = x(t_uso,:);                   % shows the x values index on t>2 years

  dummy_xmean(fu,ind,:)=nanmean(x_last);  %mean of x values (density) in the last 2 years (this is the outcome that I want to save in a matrix array) 
  % I've set de position in wich I want each outcome t be saved
 
   
      end
      
       x_mean(fu,ind,:)=dummy_xmean       % here I try to save each outcome from de loops in a matrix array in its corresponding position
end

But when I run this code, matlab gets me an error "variable x_mean can't be classifiedin a parfor. Anyone can help me with this? thank you very much beforehand

P.S: do not try to run this code, because is not complete it won't run

Answer 1

[Walter Roberson]

muvec= cat(2,fliplr(180:-deltin:inferior),180+deltin:deltin:superior);
delvec= muvec - m(1);            %vector array with delta m2
l_vec= length (muvec); 
rep=10;                          %number of replicas
x_mean=zeros (rep,l_vec,n);      %matrix array to save the output (x_mean definition)
tspan = y*T; 
x_ini=zeros (rep,n);             %vector array to save all the generated random x0%  (do not pay attention to this,
%  this concern to another part of the code)
%here I start de nested parfor:
parfor fu=1:rep
    %This dummys variables are a duplicate of the above and are used to decrease run time
    dummy_x0=x_ini;             % x_ini dummy
    this_xmean = x_mean(fu,:,:);         % x_mean dummy ATENTION!
    dummy_muvec= muvec;         % muvec dummy
    dummy_m=m                   % m dummy
   %this parfor calls each x0 in the array x_ini (using its dummy) to be used in the ode45 
    x0=dummy_x0(fu,:);
    
    %this for loop calls each mu value in the array muvec (using its dummy)
    %to be used in the ode 45
      for ind=1:l_vec
          dummy_m(2)=dummy_muvec(:,ind);
                 
                          
         
         [t, x] = ode45(@odesystem, [0, tspan], x0, [], ...
                       n, K, N, S, A, T, b, s, p, f, m, w, a, d);
                   
          t_uso = find(t>730);                   % find the indices where t>2 years
          x_last = x(t_uso,:);                   % shows the x values index on t>2 years
          this_xmean(ind,:)=nanmean(x_last);  %mean of x values (density) in the last 2 years (this is the outcome that I want to save in a matrix array) 
          % I've set de position in wich I want each outcome t be saved
      end
      
   x_mean(fu,:,:)=this_xmean       % here I try to save each outcome from de loops in a matrix array in its corresponding position
end

[Beatriz Sanchez]

hi! thank you very much for your answer... I see my error now, the problem is that everytime that the parfor loop run request the outcome from another core, so it can't run. The way you do it is wrong also, for the same reason.... What I have to do is to change the definition of dummy_xmean:

dummy_xmean= zeros (rep, l_vec, n)

that way, each iteration is independent from each other.