|
|
#ifndef TIMESERIESUTILS_H
|
|
|
#define TIMESERIESUTILS_H
|
|
|
#include "MultiComponentTimeSerie.h"
|
|
|
#include "ScalarTimeSerie.h"
|
|
|
#include "SpectrogramTimeSerie.h"
|
|
|
#include "VectorTimeSerie.h"
|
|
|
|
|
|
#include <TimeSeries.h>
|
|
|
#include <cmath>
|
|
|
#include <memory>
|
|
|
|
|
|
namespace TimeSeriesUtils
|
|
|
{
|
|
|
template<typename T> TimeSeries::ITimeSerie* copy(T input_ts)
|
|
|
{
|
|
|
if constexpr(std::is_base_of_v<TimeSeries::ITimeSerie, T>)
|
|
|
{
|
|
|
if(auto ts = dynamic_cast<VectorTimeSerie*>(input_ts))
|
|
|
{ return new VectorTimeSerie(*ts); }
|
|
|
if(auto ts = dynamic_cast<ScalarTimeSerie*>(input_ts))
|
|
|
{ return new ScalarTimeSerie(*ts); }
|
|
|
if(auto ts = dynamic_cast<MultiComponentTimeSerie*>(input_ts))
|
|
|
{ return new MultiComponentTimeSerie(*ts); }
|
|
|
if(auto ts = dynamic_cast<SpectrogramTimeSerie*>(input_ts))
|
|
|
{ return new SpectrogramTimeSerie(*ts); }
|
|
|
}
|
|
|
else
|
|
|
{
|
|
|
if(auto ts = std::dynamic_pointer_cast<VectorTimeSerie>(input_ts))
|
|
|
{ return new VectorTimeSerie(*ts); }
|
|
|
if(auto ts = std::dynamic_pointer_cast<ScalarTimeSerie>(input_ts))
|
|
|
{ return new ScalarTimeSerie(*ts); }
|
|
|
if(auto ts = std::dynamic_pointer_cast<SpectrogramTimeSerie>(input_ts))
|
|
|
{ return new SpectrogramTimeSerie(*ts); }
|
|
|
if(auto ts = std::dynamic_pointer_cast<MultiComponentTimeSerie>(input_ts))
|
|
|
{ return new MultiComponentTimeSerie(*ts); }
|
|
|
}
|
|
|
return nullptr;
|
|
|
}
|
|
|
|
|
|
struct axis_properties
|
|
|
{
|
|
|
double range;
|
|
|
double max_resolution;
|
|
|
bool is_log;
|
|
|
double min;
|
|
|
double max;
|
|
|
};
|
|
|
|
|
|
constexpr auto IsLog = true;
|
|
|
constexpr auto IsLinear = false;
|
|
|
|
|
|
constexpr auto CheckMedian = true;
|
|
|
constexpr auto DontCheckMedian = false;
|
|
|
|
|
|
template<bool is_log, bool check_median>
|
|
|
axis_properties axis_analysis(
|
|
|
typename std::conditional<is_log, std::vector<double>&,
|
|
|
const std::vector<double>&>::type axis,
|
|
|
double given_max_resolution = std::nan(""))
|
|
|
{
|
|
|
std::vector<double> axis_diff(axis.size());
|
|
|
if constexpr(is_log)
|
|
|
{
|
|
|
std::transform(std::cbegin(axis), std::cend(axis), std::begin(axis),
|
|
|
[](const auto v) { return std::log10(v); });
|
|
|
}
|
|
|
auto min = *std::begin(axis);
|
|
|
auto max = *(std::cend(axis) - 1);
|
|
|
auto range = max - min;
|
|
|
auto min_diff = given_max_resolution;
|
|
|
if(std::isnan(min_diff))
|
|
|
{
|
|
|
std::adjacent_difference(std::cbegin(axis), std::cend(axis),
|
|
|
std::begin(axis_diff));
|
|
|
min_diff =
|
|
|
(*std::min_element(std::cbegin(axis_diff) + 1, std::cend(axis_diff)));
|
|
|
if constexpr(check_median)
|
|
|
{
|
|
|
std::nth_element(std::begin(axis_diff) + 1,
|
|
|
std::begin(axis_diff) + axis_diff.size() / 2,
|
|
|
std::end(axis_diff));
|
|
|
auto median_diff = *(std::begin(axis_diff) + axis_diff.size() / 2);
|
|
|
if(median_diff > (4 * min_diff)) min_diff = median_diff;
|
|
|
}
|
|
|
}
|
|
|
|
|
|
return {range, min_diff, is_log, min, max};
|
|
|
}
|
|
|
|
|
|
} // namespace TimeSeriesUtils
|
|
|
|
|
|
#endif
|
|
|
|
