Struct MinimaxProblem 

pub struct MinimaxProblem {
    pub funcs: Vec<Box<dyn Fn(&ArrayView1<'_, f64>) -> f64 + Send + Sync>>,
}

A minimax problem: minimize the maximum over k functions.

Represents:

min_x  max_{i=0..k-1}  f_i(x)

Example

use scirs2_optimize::robust::minimax::MinimaxProblem;
use scirs2_core::ndarray::ArrayView1;

// max(f0, f1) where f0(x)=x[0]+1, f1(x)=-x[0]+2
let problem = MinimaxProblem::new(vec![
    Box::new(|x: &ArrayView1<f64>| x[0] + 1.0),
    Box::new(|x: &ArrayView1<f64>| -x[0] + 2.0),
]);

Fields

funcs: Vec<Box<dyn Fn(&ArrayView1<'_, f64>) -> f64 + Send + Sync>>

The functions f_i(x) being maximized.

Implementations

impl MinimaxProblem

pub fn new( funcs: Vec<Box<dyn Fn(&ArrayView1<'_, f64>) -> f64 + Send + Sync>>, ) -> Self

Create a new minimax problem from a collection of functions.

pub fn eval_max(&self, x: &ArrayView1<'_, f64>) -> f64

Evaluate max_i f_i(x).

pub fn eval_all(&self, x: &ArrayView1<'_, f64>) -> Vec<f64>

Evaluate all function values f_i(x).

pub fn argmax(&self, x: &ArrayView1<'_, f64>) -> usize

Return the index of the maximizing function at x.

pub fn num_funcs(&self) -> usize

Return the number of component functions.