moving_average.cpp

Go to the documentation of this file.

#define _USE_MATH_DEFINES
#include <cmath>
#include "moving_average.h"
 
namespace sablib {
 
//
// Implementation of MovingAverage() function, std::vector<double> version
//
const std::vector<double> MovingAverage(const std::vector<double> & y, const unsigned int n)
{
    Eigen::VectorXd yy = Eigen::VectorXd::Map(y.data(), y.size());
 
    Eigen::VectorXd result = MovingAverage(yy, n);
 
    std::vector<double> result_y(result.size());
    Eigen::VectorXd::Map(result_y.data(), result_y.size()) = result;
 
    return result_y;
}
 
//
// Implementation of WeightedMovingAverage() function, std::vector<double> version
//
const std::vector<double> WeightedMovingAverage(const std::vector<double> & y, const std::vector<double> & w)
{
    Eigen::VectorXd yy = Eigen::VectorXd::Map(y.data(), y.size());
    Eigen::VectorXd ww = Eigen::VectorXd::Map(w.data(), w.size());
 
    Eigen::VectorXd result = WeightedMovingAverage(yy, ww);
 
    std::vector<double> result_y(result.size());
    Eigen::VectorXd::Map(result_y.data(), result_y.size()) = result;
 
    return result_y;
}
 
//
// Implementation of GaussianKernel() function
//
const std::vector<double> GaussianKernel(const unsigned int n, const double sigma)
{
    if(n == 0) {
        throw std::invalid_argument("GaussianKernel(): n is zero.");
    }
 
    if(sigma <= 0) {
        throw std::invalid_argument("GaussianKernel(): non-positive sigma value is given.");
    }
 
    int points = 2 * n + 1;
    std::vector<double> result(points);
 
    double coeff = 1.0 / (std::sqrt(2 * M_PI) * sigma);
    auto gaussian = [&](const double xx) {
        double t = xx / sigma;
        return coeff * std::exp(-t * t / 2);
    };
 
    double sum = 0;
    for(int i = -(int)n; i <= (int)n; i++) {
        result[i + n] = gaussian(i);
        sum += result[i + n];
    }
 
    for(int i = 0; i < points; i++) {
        result[i] /= sum;
    }
 
    return result;
}
 
}; // namespace sablib