To optimize parametric model (non-linear regression)
extern crate peroxide;
use peroxide::fuga::*;
use std::collections::HashMap;
pub struct Optimizer<F>
where F: Fn(&Vec<f64>, Vec<Number>) -> Option<Vec<Number>> {
domain: Vec<f64>,
observed: Vec<f64>,
func: Box<F>,
param: Vec<Number>,
max_iter: usize,
error: f64,
method: OptMethod,
option: HashMap<OptOption, bool>,
}
new
: Declare new Optimizer. Should be mutable
set_init_param
: Input initial parameter
set_max_iter
: Set maximum number of iterations
set_method
: Set method to optimization
get_domain
: Get domain
get_error
: Root mean square error
- Optimize $y = x^n$ model with $y = x^2$ with gaussian noise.
#[macro_use]
extern crate peroxide;
use peroxide::fuga::*;
fn main() {
let normal = Normal(0f64, 0.1f64);
let normal2 = Normal(0f64, 100f64);
let mut x = seq(0., 99., 1f64);
x = zip_with(|a, b| (a + b).abs(), &x, &normal.sample(x.len()));
let mut y = x.fmap(|t| t.powi(2));
y = zip_with(|a, b| a + b, &y, &normal2.sample(y.len()));
let n_init = vec![1f64];
let data = hstack!(x.clone(), y.clone());
let mut opt = Optimizer::new(data, quad);
let p = opt.set_init_param(n_init)
.set_max_iter(50)
.set_method(LevenbergMarquardt)
.optimize();
p.print();
opt.get_error().print();
let z = quad(&x, NumberVector::from_f64_vec(p)).unwrap().to_f64_vec();
}
fn quad(x: &Vec<f64>, n: Vec<Number>) -> Option<Vec<Number>> {
Some(
x.clone().into_iter()
.map(|t| Number::from_f64(t))
.map(|t| t.pow(n[0]))
.collect()
)
}
Optimizer | Optimizer for optimization (non-linear regression)
|