Arrayfun for a particukar nested loop in GPU computing

Hi i've this situation (CPU):

X(1)= first_step;
for k=1:length_1
for p=0:min(k-1,length_2)

X(k+1)=X(k+1)+j(p+1)*X(k-p)
end
end

I want to make it more appropriate for GPU, so my idea is to consider arrayfun instead of for loop. I found it not so easy because here i have a nested loop with a dependence from the previous components. I use min() function so in this way the difference k-p is always positive.