dolysis
/
benchmark
mirror of https://github.com/google/benchmark.git
3
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

StatisticsMedian: Fix bug 

Previously, this could return the wrong result when there
was an even number of elements.

There were two `nth_element` calls.  The second call could
change elements in `[center2, end])`, which was where
`center` pointed.  Therefore, `*center` sometimes had the
wrong value after the second `nth_element` call.

Rewrite to use `max_element` instead of the second call to
`nth_element`.  This avoids modifying the vector.
This commit is contained in:
Jesse Rosenstock 2023-08-24 16:05:09 +02:00
parent 9c65aebb26
commit 6dd50bb606
1 changed files with 5 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
src/statistics.cc
View File

@ -42,13 +42,12 @@ double StatisticsMedian(const std::vector<double>& v) {
auto center = copy.begin() + v.size() / 2;
std::nth_element(copy.begin(), center, copy.end());
// did we have an odd number of samples?
// if yes, then center is the median
// it no, then we are looking for the average between center and the value
// before
// Did we have an odd number of samples? If yes, then center is the median.
// If not, then we are looking for the average between center and the value
// before. Instead of resorting, we just look for the max value before it.
// (Since `copy` is partially sorted.)
if (v.size() % 2 == 1) return *center;
auto center2 = copy.begin() + v.size() / 2 - 1;
std::nth_element(copy.begin(), center2, copy.end());
auto center2 = std::max_element(copy.begin(), center);
return (*center + *center2) / 2.0;
}