simulating in parfor and geting a vector

I'm trying to speed up my simulation by incorporating parfor, but I do not seem able to return vectors. what basically will happen is that I define a system of odes (lorenz equations) and every parfor iteration would than just have a different set of initial conditions and calculate a path of like 100 steps. I use convhull to get the volume of each new sphere and save those in a vector and I want to do some postprocessing on this matrix of vectors.

paths = 5;
o=10;
b=8/3;
r = 28;
dt=0.001;
radius = 10^-10;
volumedata = zeros(1020, 1020);
parfor i = 1:paths
    volumedataa=[];
   %random location and initializing
    p=35*rand;
    j=35*rand;
    k=35*rand;
    [x,y,z]=sphere(10);
    xpar=radius.*x(:)+p;ypar=radius.*y(:)+j;zpar=radius.*z(:)+k;
    volumeworkerdata = [];
    for j = 1:100
        %next location for all points
        xpar=xpar+(o.*(ypar-xpar)).*dt;
        ypar=ypar+(r.*xpar-ypar-xpar.*zpar).*dt;
        zpar=zpar+(xpar.*ypar-zpar.*b).*dt;
        %find volume
        v=[xpar ypar zpar];
        [tri, volume] = convhull(v);
        volumeworkerdata = [volumeworkerdata volume];

      end   
      volumedata(i,:) = volumeworkerdata;  
end

I keep getting errors and was hopeing that one of you would know how to get this working or if there is a better way to do this in parallel

Answer 1

[Edric Ellis]

It looks like your volumeworkerdata array isn't the right size to assign into a row of volumedata. I would make the following changes:

% replace:
  volumeworkerdata = [];
% with
  volumeworkerdata = zeros(1, 1020);

And also:

% replace:
  volumeworkerdata = [volumeworkerdata volume];
% with:
  volumeworkerdata(j) = volume;

It looks like your inner for loop might not have the correct bounds.