First CRAN release.
pareto_design(): finite alternatives, minimised
objectives with known bounds, epsilon-dominance, alpha shared by a union
bound over the K x m confidence sequences, boundary from
seqbench, cost and budget.initialize_pareto(), update_objectives():
one anytime-valid confidence sequence per alternative and objective,
running intersections, auditable history and stopping reason.plausible_pareto_set(): estimated frontier, plausible
Pareto set, certified optimal and certified dominated alternatives (each
with the alternative that dominates it); epsilon-ties keep one
representative.choose_next_evaluation() and pareto_run():
adaptive choice of the next alternative by confidence-box width per unit
cost, and a run loop with a user simulator.pareto_report(); S3 class
paretoinfer_state with print,
summary, tidy, glance,
autoplot / plot_uncertain_frontier().