Algorithm 3. Parallel UNWOUNDSUM.

1:  function PARALLEL_UNWOUNDSUM(m, pz, po, pi)

2:    l = len(m)

3:    sum = [] array of l zeroes

4:    for $i\leftarrow 1$ to l + 1 in parallel, do

5:     next = shuffle(mi.w, l)

6:     for $j\leftarrow l$ to 1 do

7:      wj = shuffle(mi.w, j)

8:      tmp = (next ⋅ (l − 1) + 1)/j

9:      sumi = sumi + tmp ⋅ po

10:      next = wj − tmp ⋅ (l − j) ⋅ pz/l

11:      sumi = sumi + (1 − po) ⋅ wj ⋅ l/((l − j) ⋅ pz)

12:   return sum