Iterative estimation of Sobol’ indices based on replicated designs

In the field of sensitivity analysis, Sobol’ indices are widely used to assess importance inputs a model its output. Among methods that estimate these indices, replication procedure is noteworthy for efficient cost. A practical problem how many evaluations must be performed guarantee sufficient precision on estimates. The present paper tackles this issue by rendering iterative. idea enable addition new progressively increase accuracy These done at points located in under-explored regions experimental designs, but preserving their characteristics. key feature approach construction nested space-filling designs. For estimation first-order Latin hypercube design used. closed second-order two constructions orthogonal array proposed. Regularity and uniformity properties designs studied.