PreciseSums (version 0.3)

pairwiseSum: Return an accurate floating point sum of values

Description

This was taken by NumPy and adapted for use here. It is more accurate than a standard sum, but still has numerical issues. Its main benefit is that it is about the same amount of time as a standard time with the added accuracy.

Usage

pairwiseSum(numbers)

Arguments

numbers

A vector of numbers to sum.

Value

A sum of numbers with a rounding error of O(lg n) instead of O(n).

References

https://github.com/juliantaylor/numpy/blob/b0bc01275cac04483e6df021211c1fa2ba65eaa3/numpy/core/src/umath/loops.c.src

https://github.com/numpy/numpy/pull/3685