parfor. Error: The variable eavg in a parfor cannot be classified.

I have already read other topic on Errors and parfor but i don't find the answer to my problem.

My code give me an Error: "Error: The variable eavg in a parfor cannot be classified."

The pice of code with parfor is:

eavg = zeros(size(coeff_d,2),size(coeff_c,2),size(f,1));

parfor i0 = 1:size(vett,2)
    i1 = vett(1,i0);
    i2 = vett(2,i0);
    i3 = vett(3,i0);
            if curva~=2
                if curva==4
                    z_rel = (i3-1)/(size(y,1)-1);
                    newcoeff = (-1.817*(z_rel)^2+2.776*z_rel)*0.7;
                else
                    newcoeff = coeff_c(i2);
                end
                suptemp = EsponenzialeSingola(y(i3,:), f(i3,:),...
                    coeff_d(i1), newcoeff, plottacurve);
            else
                eavg(i1,i2,i3) = EsponenzialeDoppia(y(i3,:), f(i3,:),...
                    coeff_d(i1), coeff_c(i2), plottacurve);
            end
end

the variables coeff_c, coeff_d, f, coeff_c, vett, curva, y and plottacurva are declared befor parfor-loop.

Can you help e to fix the Error?

Answer 1

[Mohammad Abouali]

your parfor is based on i0 eavg is not being sliced by that.

So it doesn't know how to slice eavg and then once it comes out of parfor it doesn't know how to combins it.

Answer 2

[Marco Castelli]

@Mohammad Abouali : Thanks to you help I solve the problem in this way:

parfor i0 = 1:size(vett,2)
    if curva~=2
        if curva==4
            z_rel = (vett(3,i0)-1)/(size(y,1)-1);
            newcoeff = (-1.817*(z_rel)^2+2.776*z_rel)*0.7;
        else
            newcoeff = coeff_c(vett(2,i0));
        end
        eavg(i0) = EsponenzialeSingola(y(vett(3,i0),:), f(vett(3,i0),:),...
            coeff_d(vett(1,i0)), newcoeff, plottacurve);
    else
        eavg(i0) = EsponenzialeDoppia(y(vett(3,i0),:), f(vett(3,i0),:),...
            coeff_d(vett(1,i0)), coeff_c(vett(2,i0)), plottacurve);
    end
end

[Mohammad Abouali]

Great. Happy that was helpful.