Function run_on_ladder 

pub fn run_on_ladder(
    problem: &MultiFidelity,
    kind: SamplerKind,
    unit_budget: u64,
    seed: u64,
) -> Result<MfReport>

Runs a searcher on DEHB’s own fidelity schedule — the DE ablation.

Trial n runs at the fidelity DEHB’s n-th job would have used (fidelity_schedule), but the configuration comes from kind (uniform random, say) instead of DE. The difference between this and run_dehb at the same unit budget is exactly what differential evolution contributes on top of the multi-fidelity schedule.

Errors

As run_dehb.