how to slice variables to use parfor

Hi,

I cannot convert this for loop into a parfor loop: I tried to have a look to the help but I cannot understand how to slice this variable to use the parfor istead of the for loop.

tot_rows = randi([10000,20000],1);
tot_cols = randi([10,500],1);

data = zeros(tot_rows, tot_cols);

for k=1:tot_cols
    idx = randi([1 tot_rows], randi([1 tot_rows]), 1); %now is generated randomly, then in real version idx is found with a criteria...
    tmp = rand(length(idx),1);%now is generated randomly, then in real version tmp is taken from txt file...
    data(idx,k) = tmp;
end

If I try to use the parfor loop I get the following error, I suppose because "data" array cannot be recalled in this way but I cannot understand how to slice it:

tot_rows = randi([10000,20000],1);
tot_cols = randi([10,500],1);

data = zeros(tot_rows, tot_cols);

parfor k=1:tot_cols
    idx = randi([1 tot_rows], randi([1 tot_rows]), 1); %now is generated randomly, then in real version idx is found with a criteria...
    tmp = rand(length(idx),1);%now is generated randomly, then in real version tmp is taken from txt file...
    data(idx,k) = tmp;
end

Error using test_for (line 6)

Error: Unable to classify the variable 'data' in the body of the parfor-loop. For more information, see Parallel for Loops in MATLAB, "Solve Variable Classification Issues in parfor-Loops".

Thanks everyone for the help! :)

Answer 1

[Matt J]

One way,

[I,J,S]=deal(cell(tot_cols,1));

parfor k=1:tot_cols
    
    idx = randi([1 tot_rows], randi([1 tot_rows]), 1); %now is generated randomly, then in real version idx is found with a criteria...
    tmp = rand(length(idx),1);%now is generated randomly, then in real version tmp is taken from txt file...
    
    [I{k} ,J{k},S{k}]=deal(idx,idx,tmp);
    J{k}(:)=k;
    
end

data = accumarray(cell2mat([I,J]), cell2mat(S),[tot_rows, tot_cols]);

[endystrike]

thx a lot Sir! :)