mirrors/phobos
1
0
Fork 0
mirror of https://github.com/dlang/phobos.git synced 2025-05-02 08:00:48 +03:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

Document use of Kahan vs. vanilla summation

Browse source
This commit is contained in:
Andrei Alexandrescu 2013-12-26 15:07:34 -08:00
parent 1564671ace
commit ed36e854ef
1 changed files with 5 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

5
std/algorithm.d
View file

@ -1028,6 +1028,11 @@ Kahan summation) algorithm.)
For floating point inputs, calculations are made in $(D real) For floating point inputs, calculations are made in $(D real)
precision for $(D real) inputs and in $(D double) precision otherwise. precision for $(D real) inputs and in $(D double) precision otherwise.
Note that these specialized summing algorithms execute more primitive operations
than vanilla summation. Therefore, if in certain cases maximum speed is required
at expense of precision, one can use $(D reduce!((a, b) => a + b)(0, r)), which
is not specialized for summation.
*/ */
auto sum(R)(R r) auto sum(R)(R r)
if (isInputRange!R && !isFloatingPoint!(ElementType!R) && !isInfinite!R) if (isInputRange!R && !isFloatingPoint!(ElementType!R) && !isInfinite!R)