1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120

use super::PositiveDefiniteKernel;
use crate::{KernelAdd, KernelError, KernelMul, ValueDifferentiable, ParamsDifferentiable};
use opensrdk_linear_algebra::Vector;
use rayon::prelude::*;
use std::{ops::Add, ops::Mul};

const PARAMS_LEN: usize = 2;

#[derive(Clone, Debug)]
pub struct RBF;

impl RBF {
    fn norm_pow(
        &self,
        params: &[f64],
        x: &Vec<f64>,
        xprime: &Vec<f64>,
    ) -> Result<f64, KernelError> {
        if params.len() != PARAMS_LEN {
            return Err(KernelError::ParametersLengthMismatch.into());
        }
        if x.len() != xprime.len() {
            return Err(KernelError::InvalidArgument.into());
        }

        let norm_pow = x
            .par_iter()
            .zip(xprime.par_iter())
            .map(|(x_i, xprime_i)| (x_i - xprime_i).powi(2))
            .sum();

        Ok(norm_pow)
    }
}

impl PositiveDefiniteKernel<Vec<f64>> for RBF {
    fn params_len(&self) -> usize {
        PARAMS_LEN
    }

    fn value(&self, params: &[f64], x: &Vec<f64>, xprime: &Vec<f64>) -> Result<f64, KernelError> {
        let norm_pow = self.norm_pow(params, x, xprime)?;

        let fx = params[0] * (-norm_pow / params[1]).exp();

        Ok(fx)
    }
}

impl<R> Add<R> for RBF
where
    R: PositiveDefiniteKernel<Vec<f64>>,
{
    type Output = KernelAdd<Self, R, Vec<f64>>;

    fn add(self, rhs: R) -> Self::Output {
        Self::Output::new(self, rhs)
    }
}

impl<R> Mul<R> for RBF
where
    R: PositiveDefiniteKernel<Vec<f64>>,
{
    type Output = KernelMul<Self, R, Vec<f64>>;

    fn mul(self, rhs: R) -> Self::Output {
        Self::Output::new(self, rhs)
    }
}

impl ValueDifferentiable<Vec<f64>> for RBF {
  fn ln_diff_value(
      &self,
      params: &[f64],
      x: &Vec<f64>,
      xprime: &Vec<f64>,
  ) -> Result<(Vec<f64>, f64), KernelError> {
      let value = &self.value(params, x, xprime).unwrap();
      let diff = (-2.0 / params[0] * (x.clone().col_mat() - xprime.clone().col_mat())).vec();
      Ok((diff, *value))
  }
}

impl ParamsDifferentiable<Vec<f64>> for RBF {
fn ln_diff_params(
    &self,
    params: &[f64],
    x: &Vec<f64>,
    xprime: &Vec<f64>,
) -> Result<(Vec<f64>, f64), KernelError> {
    let diff0 = 1.0 / params[0] ;
    let diff1 = 2.0 * params[1].powi(-2) * &self.norm_pow(params, x, xprime).unwrap();
    let diff = vec![diff0, diff1];
    let value = &self.value(params, x, xprime).unwrap();
    Ok((diff, *value))
}
}

#[cfg(test)]
mod tests {
    use crate::*;
    #[test]
    fn it_works() {
        let kernel = RBF;

        //let (func, grad) = kernel
        //    .value_with_grad(&[1.0, 1.0], &vec![1.0, 2.0, 3.0], &vec![3.0, 2.0, 1.0])
        //    .unwrap();

        //println!("{}", func);
        //println!("{:#?}", grad);

        let test_value = kernel
            .value(&[1.0, 1.0], &vec![1.0, 0.0, 0.0], &vec![0.0, 0.0, 0.0])
            .unwrap();

        assert_eq!(test_value, (-1f64).exp());
    }
}