template struct mrpt::math::RobustKernel<rkLeastSquares, T>

No robust kernel, use standard least squares: rho(r) = r^2.

#include <mrpt/math/robust_kernels.h>

template <typename T>
struct RobustKernel<rkLeastSquares, T>
{
    //
fields

    T param_sq = 1;

    //
methods

    T eval(const T r2, T& out_1st_deriv, T& out_2nd_deriv);
};

Fields

T param_sq = 1

The kernel parameter (the “threshold”) squared [Not used in this class, provided for consistency with the other classes].

Methods

T eval(const T r2, T& out_1st_deriv, T& out_2nd_deriv)

Evaluates the kernel function for the squared error r2 and returns robustified squared error and derivatives of sqrt(2*rho(r)) at this point.