The variable in a parfor cannot be classified, parfor

Hi, I have tried to convert my problem in a simple form. I know that I have to take care of the indexing in parfor loops, but I don't know how can avoid the problem that one line depends on the previous one. Please check the code and you will understand. Hope you can help me Marc

num_col = 1000;
days_simulation = 400;
SimulationPQ = nan(days_simulation,num_col);
Ergebnis = nan(days_simulation,num_col);
Ergebnis (1,:) = ones(1,num_col);

parfor kk=1:100    
    for simday = 1:400       
        SimulationPQ(simday,kk) = rand();
        Ergebnis(simday+1,kk) = Ergebnis(simday,kk) * SimulationPQ(simday,kk) ;                 
    end    
end

Answer 1

[Adam]

Something similar to this should work (it gives no M-Lint warnings at least)

num_col = 1000;
days_simulation = 400;
SimulationPQ = nan(days_simulation,num_col);
Ergebnis = nan(days_simulation,num_col);
Ergebnis (1,:) = ones(1,num_col);
parfor kk=1:100
  
  localErgebnis = Ergebnis(:,kk)
  for simday = 1:399
    SimulationPQ(simday,kk) = rand();
    localErgebnis(simday+1) = localErgebnis(simday) * SimulationPQ(simday,kk) ;
  end
  Ergebnis(:,kk) = localErgebnis;
end

except that simday+1 will go out of range, but that is just a general problem with your example, nothing to do with the parallel processing aspect.

[Marc Paul]

It works! Thank you Adam. Have a nice weekend

[Jan]

Marc: if this answer solves your problem, please accept it. Then the readers in the forum can see, that you do not need help anymore. Thanks.

[Marc Paul]

You're right. I clicked only the link in my mail to accept the answer and thought thtat's it. Sorry. Already corrected :)