Module stochastic 

Stochastic optimization via Sample-Average Approximation (SAA).

Given an objective f(x, xi) that depends on random parameters xi, SAA draws N scenarios and solves:

min_x (1/N) sum_{s=1}^{N} f(x, xi_s)

Chance constraints P{g(x, xi) <= 0} >= alpha are enforced via a smooth quadratic penalty relaxation.

Example

use numra_optim::stochastic::StochasticProblem;

let result = StochasticProblem::new(1)
    .x0(&[0.0])
    .objective(|x: &[f64], p: &[f64]| (x[0] - p[0]) * (x[0] - p[0]))
    .param_normal("xi", 5.0, 1.0)
    .n_samples(200)
    .solve()
    .unwrap();

assert!((result.x[0] - 5.0).abs() < 0.5);

Author: Moussa Leblouba Date: 9 February 2026 Modified: 2 May 2026

Structs

StochasticOptions

Options for stochastic optimization.

StochasticParam

A stochastic parameter with a sampling function.

StochasticProblem

Declarative builder for stochastic optimization problems.

StochasticResult

Result of stochastic optimization.

Functions

param_normal

Create a normally distributed stochastic parameter.

param_sampled

Create a stochastic parameter with a custom sampling closure.

param_uniform

Create a uniformly distributed stochastic parameter.