3rd-party/catch2
1
0
Fork 0
mirror of https://github.com/catchorg/Catch2.git synced 2024-11-26 07:16:10 +01:00
Code Issues Packages Projects Releases Wiki Activity

Internal linkage for outlier_variance

Browse Source
This commit is contained in:
Martin Hořeňovský 2023-05-01 00:51:43 +02:00
parent 10f0a58643
commit 51fdbedd13
No known key found for this signature in database
GPG Key ID: DE48307B8B0D381A
2 changed files with 157 additions and 143 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

118
src/catch2/benchmark/detail/catch_stats.cpp
View File

@ -21,45 +21,84 @@
#include <future> #include <future>
#endif #endif
namespace { namespace Catch {
namespace Benchmark {
using Catch::Benchmark::Detail::sample; namespace Detail {
namespace {
template <typename URng, typename Estimator> template <typename URng, typename Estimator>
static sample resample(URng& rng, unsigned int resamples, static sample
resample( URng& rng,
unsigned int resamples,
std::vector<double>::const_iterator first, std::vector<double>::const_iterator first,
std::vector<double>::const_iterator last, std::vector<double>::const_iterator last,
Estimator& estimator) { Estimator& estimator ) {
auto n = static_cast<size_t>(last - first); auto n = static_cast<size_t>( last - first );
std::uniform_int_distribution<decltype(n)> dist(0, n - 1); std::uniform_int_distribution<decltype( n )> dist( 0,
n - 1 );
sample out; sample out;
out.reserve(resamples); out.reserve( resamples );
// We allocate the vector outside the loop to avoid realloc per resample // We allocate the vector outside the loop to avoid realloc
// per resample
std::vector<double> resampled; std::vector<double> resampled;
resampled.reserve( n ); resampled.reserve( n );
for ( size_t i = 0; i < resamples; ++i ) { for ( size_t i = 0; i < resamples; ++i ) {
resampled.clear(); resampled.clear();
for ( size_t s = 0; s < n; ++s ) { for ( size_t s = 0; s < n; ++s ) {
resampled.push_back( resampled.push_back(
first[static_cast<std::ptrdiff_t>( dist( rng ) )] ); first[static_cast<std::ptrdiff_t>(
dist( rng ) )] );
} }
const auto estimate = const auto estimate =
estimator( resampled.begin(), resampled.end() ); estimator( resampled.begin(), resampled.end() );
out.push_back( estimate ); out.push_back( estimate );
} }
std::sort(out.begin(), out.end()); std::sort( out.begin(), out.end() );
return out; return out;
} }
static double outlier_variance( Estimate<double> mean,
Estimate<double> stddev,
int n ) {
double sb = stddev.point;
double mn = mean.point / n;
double mg_min = mn / 2.;
double sg = (std::min)( mg_min / 4., sb / std::sqrt( n ) );
double sg2 = sg * sg;
double sb2 = sb * sb;
double erf_inv(double x) { auto c_max = [n, mn, sb2, sg2]( double x ) -> double {
// Code accompanying the article "Approximating the erfinv function" in GPU Computing Gems, Volume 2 double k = mn - x;
double d = k * k;
double nd = n * d;
double k0 = -n * nd;
double k1 = sb2 - n * sg2 + nd;
double det = k1 * k1 - 4 * sg2 * k0;
return static_cast<int>( -2. * k0 /
( k1 + std::sqrt( det ) ) );
};
auto var_out = [n, sb2, sg2]( double c ) {
double nc = n - c;
return ( nc / n ) * ( sb2 - nc * sg2 );
};
return (std::min)( var_out( 1 ),
var_out(
(std::min)( c_max( 0. ),
c_max( mg_min ) ) ) ) /
sb2;
}
static double erf_inv( double x ) {
// Code accompanying the article "Approximating the erfinv
// function" in GPU Computing Gems, Volume 2
double w, p; double w, p;
w = -log((1.0 - x) * (1.0 + x)); w = -log( ( 1.0 - x ) * ( 1.0 + x ) );
if (w < 6.250000) { if ( w < 6.250000 ) {
w = w - 3.125000; w = w - 3.125000;
p = -3.6444120640178196996e-21; p = -3.6444120640178196996e-21;
p = -1.685059138182016589e-19 + p * w; p = -1.685059138182016589e-19 + p * w;
@ -84,8 +123,8 @@ using Catch::Benchmark::Detail::sample;
p = -0.0060336708714301490533 + p * w; p = -0.0060336708714301490533 + p * w;
p = 0.24015818242558961693 + p * w; p = 0.24015818242558961693 + p * w;
p = 1.6536545626831027356 + p * w; p = 1.6536545626831027356 + p * w;
} else if (w < 16.000000) { } else if ( w < 16.000000 ) {
w = sqrt(w) - 3.250000; w = sqrt( w ) - 3.250000;
p = 2.2137376921775787049e-09; p = 2.2137376921775787049e-09;
p = 9.0756561938885390979e-08 + p * w; p = 9.0756561938885390979e-08 + p * w;
p = -2.7517406297064545428e-07 + p * w; p = -2.7517406297064545428e-07 + p * w;
@ -106,7 +145,7 @@ using Catch::Benchmark::Detail::sample;
p = 1.0052589676941592334 + p * w; p = 1.0052589676941592334 + p * w;
p = 3.0838856104922207635 + p * w; p = 3.0838856104922207635 + p * w;
} else { } else {
w = sqrt(w) - 5.000000; w = sqrt( w ) - 5.000000;
p = -2.7109920616438573243e-11; p = -2.7109920616438573243e-11;
p = -2.5556418169965252055e-10 + p * w; p = -2.5556418169965252055e-10 + p * w;
p = 1.5076572693500548083e-09 + p * w; p = 1.5076572693500548083e-09 + p * w;
@ -128,10 +167,12 @@ using Catch::Benchmark::Detail::sample;
return p * x; return p * x;
} }
double standard_deviation(std::vector<double>::const_iterator first, static double
std::vector<double>::const_iterator last) { standard_deviation( std::vector<double>::const_iterator first,
auto m = Catch::Benchmark::Detail::mean(first, last); std::vector<double>::const_iterator last ) {
double variance = std::accumulate( first, auto m = Catch::Benchmark::Detail::mean( first, last );
double variance =
std::accumulate( first,
last, last,
0., 0.,
[m]( double a, double b ) { [m]( double a, double b ) {
@ -142,7 +183,10 @@ using Catch::Benchmark::Detail::sample;
return std::sqrt( variance ); return std::sqrt( variance );
} }
} } // namespace
} // namespace Detail
} // namespace Benchmark
} // namespace Catch
namespace Catch { namespace Catch {
namespace Benchmark { namespace Benchmark {
@ -234,34 +278,6 @@ namespace Catch {
return result; return result;
} }
double outlier_variance(Estimate<double> mean, Estimate<double> stddev, int n) {
double sb = stddev.point;
double mn = mean.point / n;
double mg_min = mn / 2.;
double sg = (std::min)(mg_min / 4., sb / std::sqrt(n));
double sg2 = sg * sg;
double sb2 = sb * sb;
auto c_max = [n, mn, sb2, sg2](double x) -> double {
double k = mn - x;
double d = k * k;
double nd = n * d;
double k0 = -n * nd;
double k1 = sb2 - n * sg2 + nd;
double det = k1 * k1 - 4 * sg2 * k0;
return static_cast<int>(-2. * k0 / (k1 + std::sqrt(det)));
};
auto var_out = [n, sb2, sg2](double c) {
double nc = n - c;
return (nc / n) * (sb2 - nc * sg2);
};
return (std::min)(var_out(1), var_out((std::min)(c_max(0.), c_max(mg_min)))) / sb2;
}
bootstrap_analysis analyse_samples(double confidence_level, bootstrap_analysis analyse_samples(double confidence_level,
unsigned int n_resamples, unsigned int n_resamples,
std::vector<double>::iterator first, std::vector<double>::iterator first,

2
src/catch2/benchmark/detail/catch_stats.hpp
View File

@ -108,8 +108,6 @@ namespace Catch {
return { point, resample[lo], resample[hi], confidence_level }; return { point, resample[lo], resample[hi], confidence_level };
} }
double outlier_variance(Estimate<double> mean, Estimate<double> stddev, int n);
struct bootstrap_analysis { struct bootstrap_analysis {
Estimate<double> mean; Estimate<double> mean;
Estimate<double> standard_deviation; Estimate<double> standard_deviation;