An adaptive design and interpolation technique for extracting highly nonlinear response surfaces from deterministic models

摘要: Response surface methodologies can reveal important features of complex computer code models. Here, we suggest experimental designs and interpolation methods for extracting nonlinear response surfaces whose roughness varies substantially over the input domain. A sequential design algorithm cuboid domains is initiated by selecting an extended corner/centre point entire domain, then updated decomposing this domain into disjoint cuboids taking corners centre these as new points. criterion used to control decomposition so that becomes space-filling coverage particularly good in parts where strongly nonlinear. Finally, model output at untried inputs predicted carefully a local neighbourhood each space fitting full quadratic polynomial data points neighbourhood. Test runs showed our automatically adapts output. Moreover, technique useful from models with two seven variables. simple modification outlined enables adequate handling non-cuboid domains.