data sampling

[UlrikHagerRoiser]

Hello, Mr. Madsen,
I am trying to use the data sampling function:
1: function DATASET(OP(.,.) : Operation, R : Range, d = 100, s = 0:25, o = 0:5)
2: x <- UNIFORM(Rlower;Rupper; i) . Sample d elements uniformly
3: k <- UNIFORM(0; 1 - 2s - o) . Sample offset. Same for interpolation and extrapolation.
4: a <- SUM(x[dk : d(k + s)]) . Create sum a from a subset of length s � d
5: b <- SUM(x[d(k + s - o) : d(k + 2s - 0)]) . Create sum b from a subset of length s � d
6: t <- OP(a,b) . Perform operation on a and b
7: return x; t

from your paper

"Measuring Arithmetic Extrapolation Performance"

The piece of pseudocode that is given in the paper seems to be flawed as the set for b seems always to be larger by the size of the overlap. I think in line 5 there should be o instead of 0.
Probably, one can omitt the k and always take a from indices 0:ds and b from d(s-o):d(2s-o), where o<=s.

Do yo have a source for this sampling function or did you just make it up?

Kind regards,

Ulrik Hager-Roiser