On Mon, Jul 26, 2010 at 6:59 PM, Gaetano Mendola <mendola@gmail.com> wrote:
On 07/26/2010 06:55 PM, Jeffrey Lee Hellrung, Jr. wrote:
Or is the proposition to replace the (naive) summation algorithm with kahan summation algorithm for only floating point types?
A possible scenario can be:
1) Leave the naive summation algorith as it is now 2) Create (with a typedef?) fast_sum that is the same of sum (naive) 2) Rename the sum_kahan as accurate_sum implemented as sum (naive) 3) Specialize the accurate_sum for floating points types implemented as the real kahan's algorithm
however this can lead to have someone using accurate_sum for integral type and find it out is not accurate at all !
wouldn't the native sum be accurate (actually exact) for integral types as long as the sum doesn't overflow? -- gpd