Skip to main content

superforecaster

wm_simulation::pce

Function superforecaster 

Source 
pub fn superforecaster<F>(
    bounds: &[(f64, f64)],
    fitness: F,
    n_initial: usize,
    n_bo_iterations: usize,
    pce_degree: usize,
    seed: u64,
) -> SuperforecasterResult
where
    F: Fn(&[f64]) -> f64 + Clone,
Expand description

Run the full superforecaster pipeline.

  1. LHS sample n_initial points in the box and evaluate the fitness.
  2. Fit a PCE (degree pce_degree) → Sobol’ indices (sensitivity).
  3. Warm-start Bayesian optimization with the LHS evaluations and refine for n_bo_iterations iterations.