Improving Floating Point Accuracy: A Look at Sums

Measuring The Error of Floating Point Programs