Struct StochasticProblem 

pub struct StochasticProblem<S: Scalar> { /* private fields */ }

Declarative builder for stochastic optimization problems.

Build a problem with the fluent API, then call .solve() to perform Sample-Average Approximation (SAA) optimization.

Implementations

impl<S: Scalar> StochasticProblem<S>

pub fn new(n: usize) -> Self

Create a new stochastic optimization problem with n decision variables.

pub fn x0(self, x0: &[S]) -> Self

Set the initial point.

pub fn bounds(self, i: usize, lo_hi: (S, S)) -> Self

Set bounds for variable i.

pub fn all_bounds(self, bounds: &[(S, S)]) -> Self

Set bounds for all variables at once.

pub fn objective<F>(self, f: F) -> Self

where F: Fn(&[S], &[S]) -> S + Send + Sync + 'static,

Set the parameterized objective function f(x, params).

The solver minimizes the expected value E[f(x, xi)] over the stochastic parameters.

pub fn constraint_det_ineq<F>(self, f: F) -> Self

where F: Fn(&[S]) -> S + Send + Sync + 'static,

Add a deterministic inequality constraint g(x) <= 0.

pub fn constraint_det_eq<F>(self, f: F) -> Self

where F: Fn(&[S]) -> S + Send + Sync + 'static,

Add a deterministic equality constraint h(x) = 0.

pub fn chance_constraint<F>(self, f: F, probability: S) -> Self

where F: Fn(&[S], &[S]) -> S + Send + Sync + 'static,

Add a chance constraint: P{g(x, xi) <= 0} >= probability.

The constraint function g(x, params) should return a value where g <= 0 means the constraint is satisfied.

pub fn param(self, p: StochasticParam<S>) -> Self

Add a stochastic parameter directly.

pub fn param_normal(self, name: &str, mean: S, std: S) -> Self

Add a normally distributed stochastic parameter.

pub fn param_uniform(self, name: &str, lo: S, hi: S) -> Self

Add a uniformly distributed stochastic parameter.

pub fn n_samples(self, n: usize) -> Self

Set the number of SAA samples.

pub fn seed(self, s: u64) -> Self

Set the random seed.

pub fn max_iter(self, n: usize) -> Self

Set the maximum number of optimizer iterations.

pub fn minimize_cvar(self, alpha: S) -> Self

Minimize CVaR (Conditional Value at Risk) at confidence level alpha instead of the expected value.

Uses the Rockafellar-Uryasev reformulation:

CVaR_alpha(X) = min_t { t + 1/(1-alpha) * E[max(0, X - t)] }

An auxiliary variable t (VaR estimate) is appended to the decision vector. Requires alpha in (0, 1), typically 0.9 or 0.95.

impl<S: Scalar + SimpleEntity + Conjugate<Canonical = S> + ComplexField> StochasticProblem<S>

pub fn solve(self) -> Result<StochasticResult<S>, OptimError>

Solve the stochastic optimization problem using SAA.

1. Generates n_samples scenarios by sampling all stochastic parameters.
2. Builds a deterministic SAA problem: min (1/N) sum f(x, xi_s).
3. Chance constraints are enforced via smooth quadratic penalty.
4. Solves and evaluates constraint satisfaction at the optimal point.